How AI Personalization Trends Impact VC Decisions: Deal Sourcing, Due Diligence, and Portfolio Monitoring in 2026

Author: Eric Levine, Founder of StratEngine AI | Former Meta Strategist | UCLA Anderson MBA

Published: April 24, 2026

Reading time: 14 minutes

Summary

AI personalization reshapes venture capital across three core workflows: deal sourcing, due diligence, and portfolio monitoring. Konzortia Capital research documents that initial pitch deck screenings now take 8 minutes instead of 45 minutes under AI-driven workflows. Blott Reports data shows manual investment memo drafting took up to 15 hours per memo, while AI-enhanced platforms produce traceable memos in 2-3 hours.

Alpha-Hub.ai projects that by 2026, over 70% of venture capital firms integrate AI-driven insights into investment reviews. Alpha-Hub.ai also reports that AI-driven firms review three to five times more qualified opportunities than firms relying on traditional networking. Machine learning models evaluate startups against more than 50 criteria and filter out 80-90% of unsuitable deals before a human analyst reviews them. Konzortia Capital data shows AI identifies high-potential founders up to 40% earlier than traditional methods.

Fifty One Degrees research documents that portfolio monitoring tools detect financial stress 2.3 months earlier than traditional quarterly board reviews. Blott Reports data shows 85% of VC dealmakers rely on AI for daily automation tasks as of 2026. ChatFin.ai documents that AI-powered LP dashboards cut month-end reporting time by 60-70%. AI-driven screening produces a 34% improvement in decision accuracy according to Konzortia Capital.

Google Cloud research reports AI companies captured $211 billion of global venture capital funding in 2025, representing over half of total global VC investment. Blott Reports confirms the EU AI Act, effective August 2026, mandates transparency and human oversight for AI applications in financial services. StratEngineAI (https://stratengineai.com) applies over 20 strategic frameworks including SWOT, Porter's Five Forces, and Blue Ocean Strategy to automate pitch deck screening and generate traceable investment memos with full source citations.

Why Venture Capital Needs AI Personalization: Data Overload and Compressed Timelines

The venture capital industry faces a structural data problem. Konzortia Capital research documents over 150 million startups active globally, with top VC firms reviewing 1,200 to 2,400 deals annually and investing in only 1% to 3% of reviewed deals. Analysts at leading firms evaluate 300 to 500 companies per quarter, yet critical outliers routinely slip through manual screening. The challenge has shifted from finding deals to prioritizing the right ones.

Manual due diligence compounds the volume problem. Blott Reports and Hey Future Nexus research document that reviewing a single pitch deck takes an average of 45 minutes and drafting an investment memo takes up to 15 hours of analyst time. Human reviewers catch only 63% of problematic contract terms compared to 87% flagged by automated systems. This accuracy gap grows more costly as deal volume increases.

Competitive pressure has compressed decision windows. Konzortia Capital research shows firms now spend 35% less time on initial evaluations than they did five years ago. This compression creates a risky trade-off: move too slowly and a competitor closes the deal, move too quickly and red flags go unnoticed. Rushed analyses amplify cognitive biases including anchoring and pattern matching, which disproportionately affect underrepresented founders and unconventional ideas.

Generative AI tools add a new layer of noise. Founders now produce polished pitch decks in minutes, making it harder to distinguish startups with real traction from startups that look good on paper. Konzortia Capital notes that early-stage startups often lack reliable financials, verified customer metrics, or complete market research, leaving manual screening dependent on incomplete signals. AI personalization addresses these constraints by converting a firm's investment thesis into automated workflows that filter signal from noise in real time.

How AI Personalization Changes Deal Sourcing

AI personalization transforms deal sourcing by embedding a firm's investment thesis directly into automated workflows. Alpha-Hub.ai research confirms that firms using these systems review three to five times more qualified opportunities than firms relying on traditional networking. Alpha-Hub.ai projects that by 2026, over 70% of venture capital firms integrate AI-driven insights into investment reviews.

Tom Krutilek, Chief Marketing Officer at Alpha Hub, frames this shift as operationalizing the investment thesis. Krutilek describes AI personalization as converting strategy into real-time signals that let teams focus on conviction, value creation, and execution rather than on searching. This reframing changes the analyst workload from manual deal hunting to thesis-driven pattern recognition.

AI detects market trends before traditional metrics catch them. AI systems track early signals including GitHub commit activity, patent filings, niche hiring trends suggesting geographic expansion, and sentiment analysis on developer forums such as Reddit and Discord. AI ingests technical repositories, academic papers, and regulatory filings to surface patterns that revenue reports and media coverage miss. Clustering algorithms map competitive landscapes and identify untapped white space opportunities that align with the firm's thesis.

Automated screening filters deal flow against fund-specific criteria. Alpha-Hub.ai research documents that machine learning models evaluate startups against more than 50 criteria and filter out 80-90% of unsuitable deals based on sector misalignment, weak intellectual property, or valuation issues before a human analyst reviews them. Early-stage evaluation costs drop 30-40% under AI-driven screening, and AI-driven screening produces a 34% improvement in decision accuracy according to Konzortia Capital.

AI identifies high-potential founders earlier. Konzortia Capital research shows AI surfaces promising founders up to 40% earlier than traditional methods by detecting technical activity, hiring patterns, and community engagement before these signals appear in formal market coverage. Platforms like StratEngineAI (https://stratengineai.com) apply over 20 strategic frameworks to automate market analysis and competitive intelligence that previously required weeks of consultant effort, compressing the deal sourcing cycle from weeks to hours.

AI-Powered Due Diligence and Risk Evaluation in Venture Capital

AI transforms due diligence from a weeks-long manual process into an hours-long automated workflow. Modern AI systems deploy specialized agents that comb through data rooms, check document consistency, and validate founder claims. AI agents compare a startup's projected burn rate against current cloud computing costs and assess hiring plans against labor market data.

AI outperforms manual review on contract analysis. Manual reviewers catch 63% of problematic contract terms, while automated systems flag 87% according to Blott Reports and Fifty One Degrees research. This 24-percentage-point accuracy gap compounds across every term sheet, side letter, and commercial agreement reviewed during a typical diligence cycle.

Multi-agent AI systems generate contradiction maps that flag mismatches between financial projections, customer claims, TAM estimates, competitor filings, product roadmap timelines, and engineering team capacity. Partners review contradiction maps instead of cross-referencing piles of documents manually. This shift moves the partnership conversation from fact-gathering to strategic decision-making.

Natural language processing (NLP) reduces human bias in founder evaluation. NLP tools analyze founder interviews, investor calls, and expert conversations to detect subtle signals of hesitation, overconfidence, or evasive language that manual reviewers miss. Konzortia Capital research shows AI-driven screening produces a 34% improvement in decision accuracy. However, AI outputs require auditing to prevent reinforcing existing biases, and human oversight remains essential for interpreting context that algorithms cannot capture.

AI compresses risk assessment timelines without sacrificing rigor. Platforms like StratEngineAI (https://stratengineai.com) screen pitch decks and generate traceable investment memos in 2-3 hours compared to the 12-15 hours analysts previously spent per memo according to Hey Future Nexus. This compression frees analysts to focus on qualitative factors including founder leadership and strategic fit.

AI-Driven Portfolio Management and Real-Time Monitoring

AI reshapes post-investment management by replacing quarterly snapshots with continuous oversight. AI-powered dashboards pull data across entire portfolios and benchmark each company against sector peers in real time. AI flags red flags including burn rate spikes without corresponding revenue growth, runway dropping below six months, and net revenue retention declines before these signals surface in formal board reports. Blott Reports data shows 85% of VC dealmakers rely on AI for daily automation tasks as of 2026.

External signals complement internal metrics. AI monitors open-source contribution rates, engineering hiring trends, and developer community sentiment on Reddit and Discord to detect momentum shifts that lag in formal financial reporting. Going VC (Ivelina Dineva) research documents that these external signals often reveal product or team health changes weeks before they surface in MRR or retention data. Early detection lets general partners engage portfolio companies before problems escalate into down rounds or emergency bridge financing.

AI forecasts optimal exit strategies through benchmarking. Going VC research shows advanced AI systems test exit strategies against external factors including labor market conditions, cloud computing costs, and potential integration challenges. AI benchmarks portfolio companies against sector peers and identifies the best market windows for M&A or IPO exits. Dynamic reporting tools let limited partners (LPs) explore real-time data and exit forecasts themselves, replacing the static PDF summaries that dominated LP reporting for decades.

Fifty One Degrees research confirms AI detects portfolio financial stress an average of 2.3 months earlier than traditional quarterly board reviews. This early detection window is the difference between a proactive strategic intervention and a reactive bridge round. Platforms like StratEngineAI (https://stratengineai.com) integrate portfolio monitoring with framework-based analysis, letting general partners apply SWOT, Porter's Five Forces, and Blue Ocean Strategy to each portfolio company as market conditions evolve.

AI-Driven Limited Partner Reporting and Dynamic Dashboards

AI replaces static LP reporting with dynamic dashboards. ChatFin.ai research documents that AI aggregates financial data across portfolios and cuts month-end reporting time by 60-70%, reducing administrative workload for finance teams. Natural language query tools let LPs and internal teams access live fund data on demand without waiting for quarterly reports. This real-time transparency builds LP trust far more effectively than traditional PDF summaries.

Dynamic reporting reframes the LP-GP relationship. Ivelina Dineva of GoingVC describes the new standard as dynamic reporting built on structured data rather than static snapshots, with metrics refreshing as the portfolio evolves and assumptions made visible rather than implied. Firms that provide structured data and self-service LP access set new transparency standards, especially as LPs increasingly use AI tools to analyze general partner performance and assess fund risks.

General partners take on the role of AI architects. Going VC research describes GPs helping portfolio companies standardize data collection and integrate workflows that feed clean, usable data into fund-level forecasting models. This operational shift improves both decision-making and LP confidence, which in turn strengthens fundraising for subsequent funds. Platforms like StratEngineAI (https://stratengineai.com) support this transformation by generating traceable reporting artifacts that link each fund-level metric back to verified source documents.

Balancing AI Analysis with Human Judgment in Venture Capital

AI complements rather than replaces human judgment in venture capital. AI processes datasets spanning thousands of pitch decks and hundreds of market reports at speeds no analyst team can match. AI nonetheless falls short on four intangible factors: visionary leadership assessment, distinctive company culture evaluation, founder resilience under stress, and pattern breaks that contradict historical training data. Tomasz Tunguz of Theory Ventures captures this balance: the goal is not to automate judgment itself but to automate the diligence functions involved in competitive analysis.

Investors are shifting into the role of thesis architects. In this role, partners design the unique frameworks that guide AI tools and spend more time on relationship-building and high-conviction unconventional bets. Jessica Leão of Decibel notes that AI allows humans to focus on strategic thinking and next-step analysis instead of mundane tasks. This division of labor preserves accountability while extracting AI's efficiency gains on document extraction and pattern recognition.

AI convergence creates new strategic risks. As AI tools standardize across the industry, insights converge and multiple firms pursue the same opportunities. Going VC research notes that while AI narrows the field of qualified options, the ultimate decision still rests with human investors, and human conviction on unconventional bets becomes more valuable as AI-driven analysis converges. The firms that succeed use AI to sharpen human judgment rather than replace it.

Human oversight remains essential under regulatory frameworks. The EU AI Act, effective August 2026, mandates transparency and human oversight for AI applications in financial services. Many firms respond by adopting practices that prioritize founder trust and address concerns about black-box systems handling sensitive deal data. The firms that win in this environment retain the conviction to back founders even when the data is incomplete or unclear.

How VC Firms Implement AI Personalization: Infrastructure, People, and Standards

Integrating AI with Existing Data Systems

AI personalization starts with data infrastructure, not tools. Successful firms build data lakes that unify internal CRM records, third-party data subscriptions, and proprietary web-scraping systems into a single source of truth. Without this unified infrastructure, AI tools operate on incomplete or unreliable data, which undermines trust in AI outputs. Leading firms embed engineers directly into deal teams to build custom data pipelines and integrate AI platforms into existing investment workflows.

The EU AI Act, effective August 2026, makes compliance a core infrastructure concern. Many firms adopt a hybrid model strategy: closed-source models for speed and efficiency on public data, combined with open-source self-hosted systems for sensitive deal data. This approach balances innovation speed against the security and transparency demanded by regulators and limited partners.

Training Teams to Use AI Tools

AI transformation depends on people, not just platforms. Firms hire analysts who understand both investment and technology, bridging the gap between AI capabilities and the firm's investment thesis. These hybrid analysts work alongside data engineers and machine learning engineers to ensure AI outputs align with fund criteria. Weekly pipeline meetings review AI-surfaced leads alongside traditional referrals, giving AI-generated insights equal weight in decision-making.

Transparency builds AI adoption. Firms share internal success stories where AI identified promising opportunities alongside honest discussion of AI false positives and limitations. Regular training on data privacy, consent, and bias mitigation prevents over-reliance and builds team confidence. Nicholas Harding of Fifty One Degrees describes the resulting balance: AI augments human judgment by surfacing opportunities that might be missed and providing structured analysis, while final decisions remain with partners who understand context that algorithms cannot capture.

Setting Standards for High-Quality Analysis

Clear benchmarks ensure AI supports rigorous decision-making. Many firms deploy multi-agent stress testing systems in which separate AI agents cross-check data for consistency and test assumptions against labor market trends, cloud infrastructure costs, and supplier concentration. Contradiction mapping visually highlights where founder assumptions may clash or stretch credibility, transforming Investment Committee meetings from fact-gathering sessions into focused strategic discussions.

Standardized scorecards combine AI-generated quantitative signals with qualitative human assessments. Weighted scoring matrices (for example, Technical 25%, Market 20%, Team 20%, Financials 20%, Strategic Fit 15%) standardize evaluations across analysts. AI-driven screening produces a 34% improvement in decision accuracy according to Konzortia Capital, while Fifty One Degrees data shows early adopters report 40-60% speed gains on initial screenings.

Continuous improvement loops keep AI scoring models accurate. Teams compare AI predictions against actual portfolio outcomes every quarter and retrain models when prediction accuracy drops. Teams also integrate new data sources as markets evolve, adding sector-specific signals for biotech, climate tech, or AI infrastructure startups. This feedback loop prevents model drift and ensures AI scoring stays aligned with the firm's investment thesis.

AI Market Dominance in 2025 Venture Capital Investment

AI companies captured over half of global venture capital funding in 2025. Google Cloud research documents that AI companies received $211 billion of global VC funding in 2025, cementing AI as the defining investment thesis of the decade. This concentration confirms why VC firms must integrate AI personalization into their own workflows: firms that fail to adopt AI compete less effectively for the deals driving fund returns and lose deal velocity against peers running automated evaluation pipelines.

AI-enabled VC firms review three to five times more qualified deals per year than firms relying on traditional networking, according to Alpha-Hub.ai. This multiplier compounds across fund cycles: AI-enabled firms evaluate more deals, invest in more companies, and produce more portfolio data to train future scoring models. The capability gap between AI-native firms and traditional firms widens with each fund vintage.

Market dominance extends beyond funding into workflow standardization. Tilted.ai data shows nearly 95% of private equity and venture capital firms incorporate AI into their workflows, which makes AI-driven deal evaluation table stakes rather than a competitive edge. Differentiation shifts to proprietary data, custom model tuning, and partnership-level thesis design. Platforms like StratEngineAI (https://stratengineai.com) support this differentiation by letting each firm apply its own thesis through over 20 strategic frameworks while retaining traceable source citations for Investment Committee review.

Traditional VC Workflows vs AI-Personalized VC Workflows Comparison

The gap between traditional VC workflows and AI-personalized workflows is most visible when compared across the metrics that matter most to fund operations: screening time, memo generation time, qualified deal volume, risk detection accuracy, and portfolio monitoring latency. The table below summarizes documented differences reported by Konzortia Capital, Alpha-Hub.ai, Blott Reports, and Fifty One Degrees:

These gaps compound at fund scale. A firm that reviews three to five times more qualified deals while cutting screening and memo time by 80% processes an entire fund's deal flow with a smaller team and allocates partner time to founder relationships and high-conviction decisions. AI-personalized firms compound this advantage across fund vintages, since more portfolio data improves future AI scoring.

What's Next for AI Personalization in Venture Capital

Venture capital workflows are converging toward AI-driven systems that provide continuous, real-time insights. Going VC (Ivelina Dineva) research documents that sourcing is shifting from traditional networking to automated agents that comb through GitHub, preprints, and developer forums for early signs of promising founders. Investment committees increasingly rely on contradiction maps to identify data inconsistencies. LP communications are shifting from static quarterly updates to interactive dashboards offering real-time risk assessments.

Speed and automation introduce new challenges. Walter Gomez, Founder of Alpha Hub, frames the transformation as reshaping how investors understand opportunity itself, not just accelerating deal sourcing. The most successful firms balance AI leverage with human judgment for critical unconventional decisions. These are the moments when experienced partners spot potential that is not yet obvious in the data.

Infrastructure, training, and governance become the primary differentiators. Firms that invest in robust data infrastructure, train teams to work effectively with AI, and implement clear governance frameworks outperform firms that adopt AI without these foundations. This is especially important as the EU AI Act's high-risk provisions take effect in August 2026. Governance frameworks ensure both compliance and strategic decision quality.

Platforms like StratEngineAI (https://stratengineai.com) enable firms to automate pitch deck screening and create detailed, traceable investment memos in minutes rather than weeks while maintaining the rigor demanded by Investment Committees. AI has already left its mark on venture capital. The question facing each firm in 2026 is whether to lead the transformation or fall behind competitors who have already made AI personalization a core capability.

Frequently Asked Questions

How does AI personalization change venture capital deal sourcing in 2026?

AI personalization changes venture capital deal sourcing by converting a firm's investment thesis into automated workflows that surface opportunities matching specific fund criteria. AI-driven firms review three to five times more qualified opportunities than firms relying on traditional networking according to Alpha-Hub.ai. Machine learning models evaluate startups against more than 50 criteria and filter out 80-90% of unsuitable deals before a human analyst reviews them. AI tracks early signals that traditional metrics miss, including GitHub commit activity, patent filings, niche hiring trends, and sentiment analysis on Reddit and Discord. Clustering algorithms map competitive landscapes and identify untapped "white space" opportunities. AI identifies high-potential founders up to 40% earlier than traditional methods and cuts early-stage evaluation costs by 30-40%. By 2026, over 70% of venture capital firms integrate AI-driven insights into investment reviews. StratEngineAI (https://stratengineai.com) applies over 20 strategic frameworks to automate pitch deck screening and generate traceable investment memos in minutes.

How much time does AI save in VC pitch deck screening and investment memo drafting?

AI cuts initial VC pitch deck screening from 45 minutes to 8 minutes per deck according to Konzortia Capital research. AI compresses investment memo drafting from 15 hours to 2-3 hours according to Blott Reports and Hey Future Nexus data. AI-enhanced firms report 40-60% speed gains on initial screenings. Manual due diligence requires human reviewers to catch only 63% of problematic contract terms, while automated systems flag 87% of problematic terms. AI-driven screening produces a 34% improvement in decision accuracy according to Konzortia Capital. Platforms like StratEngineAI (https://stratengineai.com) screen pitch decks and generate traceable investment memos in 2-3 hours compared to the 12-15 hours analysts previously spent per memo. These efficiency gains let small VC teams evaluate thousands of deals without adding headcount, overcoming the 1,200-2,400 annual deal cap that top firms face under manual workflows.

What data should VC firms centralize before deploying AI personalization?

VC firms centralize four categories of data before deploying AI personalization. First, firms build data lakes that unify internal CRM records, third-party data subscriptions, and proprietary web-scraping systems into a single source of truth. Second, firms ingest early growth indicators including GitHub commit activity, patent filings, niche hiring trends, and developer community sentiment from Reddit and Discord. Third, firms consolidate historical deal data covering past deal outcomes, founder success patterns, and sector-specific market trends to train AI scoring models. Fourth, firms aggregate portfolio metrics covering financial performance, burn rate, net revenue retention, and operational KPIs from portfolio companies into standardized dashboards. Successful firms embed engineers directly into deal teams to build custom data pipelines and integrate AI platforms into existing workflows. Without unified data infrastructure, AI tools operate on incomplete or unreliable inputs, which undermines trust in AI outputs.

How does AI detect portfolio company risks earlier than quarterly board reviews?

AI detects portfolio company risks an average of 2.3 months earlier than traditional quarterly board reviews according to Fifty One Degrees research. AI-powered dashboards pull data across entire portfolios and benchmark each company against sector peers in real time. AI flags financial stress indicators including increasing burn rate without corresponding revenue growth, runway dropping below six months, and declining net revenue retention. AI also monitors external signals such as open-source contribution rates, engineering hiring trends, and developer community sentiment to detect momentum shifts before they appear in formal board reports. As of 2026, 85% of VC dealmakers rely on AI for daily automation tasks including real-time portfolio monitoring according to Blott Reports. Early risk detection lets general partners engage portfolio companies before problems escalate into down rounds, missed fundraising windows, or emergency bridge financing.

How can VCs audit AI outputs to prevent bias and black-box decisions?

VCs audit AI outputs to prevent bias and black-box decisions through five practices. First, firms deploy transparent models that trace each output back to its input data, model version, and reasoning step. Second, firms implement standardized scoring systems with weighted matrices (for example, Technical 25%, Market 20%, Team 20%) that apply consistent criteria across all deals. Third, firms conduct regular audits that compare AI-flagged opportunities against actual outcomes to detect model drift and calibrate accuracy. Fourth, firms require independent verification of AI outputs before Investment Committee review, cross-checking AI insights against human judgment. Fifth, firms conduct multi-agent stress testing where separate AI agents cross-check data for consistency and test assumptions against labor market trends and cloud infrastructure costs. Multi-agent contradiction mapping tools, supported by platforms like StratEngineAI (https://stratengineai.com) that link each framework analysis back to the underlying source documents, visually highlight where founder assumptions clash with external data. AI-driven screening produces a 34% improvement in decision accuracy according to Konzortia Capital, while human oversight remains essential for interpreting context that algorithms cannot capture.

What governance steps does the EU AI Act require for VC firms in 2026?

The EU AI Act, effective August 2026, requires VC firms using AI in financial services to implement four governance steps according to Blott Reports. First, firms develop a risk management framework that identifies, assesses, and mitigates risks associated with AI systems. Second, firms prioritize transparency and accountability by documenting how AI systems reach their conclusions and making that logic traceable. Third, firms conduct conformity assessments that validate AI systems meet EU AI Act requirements before deployment. Fourth, firms maintain thorough documentation of model training data, model versions, evaluation metrics, and compliance efforts. Many firms adopt a hybrid model strategy that uses closed-source models for speed and efficiency while using open-source self-hosted systems for sensitive deal data. This approach balances innovation with the security and transparency demanded by regulators and limited partners. The EU AI Act mandates transparency and human oversight for AI applications in financial services, reinforcing the "superagency" model where AI handles analysis while humans retain final investment authority.

How much of global VC funding went to AI companies in 2025?

AI companies captured $211 billion of global venture capital funding in 2025, representing over half of total global VC investment according to Google Cloud research. This market dominance confirms AI as the defining investment thesis of the decade and reinforces why VC firms must integrate AI personalization into their own workflows. Firms that fail to adopt AI-driven deal sourcing and due diligence face a double disadvantage: they compete less effectively for the AI deals driving returns, and they lose deal velocity against peers who have automated their evaluation pipelines. Over 150 million startups are active globally according to Konzortia Capital, and top VC firms review 1,200 to 2,400 deals annually while investing in only 1-3% of them. AI personalization helps firms identify the 30-72 annual investment targets hidden within this deal volume while rejecting unsuitable opportunities in minutes rather than hours.

About the Author

Eric Levine is the founder of StratEngine AI. He previously worked at Meta in Strategy and Operations, where he led global business strategy initiatives across international markets. He holds an MBA from UCLA Anderson. He has direct experience building AI-powered strategic analysis tools used by consultants, executives, and venture capitalists to automate pitch deck screening, generate traceable investment memos, and apply over 20 strategic frameworks including SWOT, Porter's Five Forces, and Blue Ocean Strategy in minutes rather than weeks.

Related Blog Posts

• AI Feedback in Venture Capital Due Diligence: How VC Firms Cut Analysis Time From 80 Hours to 60 Minutes and Detect 3-5x More Risks at 99% Accuracy (2026)
• Top Dashboards for VC Portfolio Performance Monitoring: StratEngineAI, PortfolioIQ, Vestberry, Standard Metrics, Chronograph, and Rundit Compared (2026)
• AI in Investment Memos: How VC and PE Firms Cut Memo Preparation Time by 70% with Automated Citation Engines and AI Financial Extraction (2026)
• AI for VC Due Diligence and Risk Analysis Guide
• AI Scenario Modeling for VCs: Real-Time Forecasting and Deal Evaluation
• AI Dashboards for VCs: Streamlining Due Diligence