Your deals already generate the data. Now you need AI systems that use it.
Marrett Advisory brings AI engineering and modern cloud architecture to boutique investment banks, private credit funds, and private equity firms. The focus is simple: build proprietary systems that turn CIMs, QoEs, models, and deal notes into structured intelligence — powering origination, counterparty matching, underwriting, portfolio monitoring, and analytics that firms can’t get from generic tools.
Most deal teams are drowning in documents and underutilized data. Whether it’s sourcing, screening, underwriting, IC prep, or monitoring, there are high-leverage workflows in every stage of your lifecycle that AI can support — if the systems are designed around how your team actually works.
Private & Secure: AI systems deployed into your firm’s own cloud environment. You own the code, the data, and the IP.
Models & Datarooms
(Firm-Specific)
Pitch & Portfolio Views
Where AI Engineering Supports Your Team
I don’t sell a platform. I build focused AI systems around your actual process — from origination and screening, through counterparty matching and underwriting, into portfolio monitoring and recurring materials.
Structure the front end of your pipeline so the right opportunities get time and the wrong ones drop out quickly.
- Automated extraction from teasers, CIMs, and inbound decks into structured fields.
- Initial screens based on size, sector, structure, leverage, and ownership narrative.
- AI-driven deal summaries and “fit” flags aligned with your mandate and strategy.
Use AI to match opportunities to the right capital providers, buyers, co-investors, or acquirers — not just lenders.
- Strict quantitative filters (EBITDA, equity check size, leverage, sector, structure) across your internal counterparty database.
- Semantic matching between deal notes and counterparty mandates using embeddings and RAG.
- AI-generated tiers and rationales that can be used in internal lists, outreach strategies, or IC materials.
Help the back half of the lifecycle — tracking, monitoring, and reporting — keep up with the front half.
- Dashboards and internal data stores for positions, covenants, and performance metrics.
- Automated production of recurring reporting inputs from portfolio company data and statements.
- Lightweight tools for tracking follow-ups, amendments, and documentation obligations.
CIM, Pitch Book & IC Memo Automation
Beyond analytics, AI engineering can handle the “last mile” of deal work: turning structured inputs into client-ready materials. The objective is not to replace judgment, but to give your team a 70–90% draft so they can focus on nuance rather than formatting.
Generate draft CIMs and pitch materials directly from deal inputs, notes, and financials.
- AI-drafted Executive Summaries, Company Overviews, Industry sections, and Investment Highlights.
- Automatic charts and tables from financials (revenue, EBITDA, margins, customer concentration).
- Reusable templates that match your firm’s structure, tone, and branding.
Turn CIMs, QoEs, call notes, and models into internal-ready investment materials.
- First-draft IC memos aligned to your firm’s sections and language.
- AI-generated risk / mitigant lists and structure commentary based on your criteria.
- Consistency across deals: no reinventing the wheel every time a memo is prepared.
Deal Intelligence from the Ground Up
Once your workflows run through a common AI and data layer, you can start to see the entire franchise differently. The same systems that screen deals and generate materials can power analytics on origination quality, execution, counterparties, IC behavior, and portfolio health.
Understand where deals are really coming from and which sources are worth compounding.
- Funnel views by source, sector, size band, and stage (screening → NDAs → IOI / LOI → Close).
- Conversion rates and time-in-stage to highlight bottlenecks and capacity constraints.
- Origination quality by banker, sponsor, intermediary, or proprietary channel.
Use match results and outcomes to sharpen outreach and structure decisions over time.
- Hit rates by lender, investor, or acquirer: NDA, IOI, term sheet, and closed transactions.
- Response speed and reliability by sector, size, and capital structure.
- Pricing and structural benchmarks: spreads, leverage, covenants, and terms by deal profile.
Tie internal decision-making and portfolio outcomes back to the underlying data and workflows.
- Analytics on IC outcomes and the risk factors that consistently drive passes or approvals.
- Portfolio health views: leverage, DSCR, and covenant headroom vs underwrite case (where data is available).
- Productivity metrics: time saved on CIMs, memos, and matching — with clear ROI to show partners.
Who I Work With
The work is best suited for lean, serious teams who handle real volume but don’t have a dedicated AI engineering group to build internal tools.
- • Lower and middle-market M&A and capital advisory shops.
- • Sector-focused boutiques with repeatable deal patterns.
- • Independent advisory teams that need better internal infrastructure.
- • Direct lending, special situations, and private credit funds.
- • Lower middle-market PE, independent sponsors, and search funds.
- • Family offices and hybrid capital providers with deal flow but limited tech bandwidth.
Security, control, and fit to your process matter more than glossy interfaces.
- You retain ownership of the code, configuration, and data.
- No need to expose confidential deal materials to generic third-party tools.
- AI systems designed around your underwriting approach and investment style, not a generic “AI product.”
How an AI Engagement Works
The structure is intentionally simple: pick one workflow, prove it out on real deals, and only then decide if it should grow. Pricing is scoped to your firm’s size and the time/value equation for the team.
We identify one repeatable workflow that slows the team down — for example: initial CIM screening, counterparty matching, pitch/CIM drafting, or IC memo preparation.
- Short call to understand how your team currently handles it.
- Review of sample (redacted) documents, models, and outputs.
I build a working AI prototype that uses your structure, terminology, and constraints — not a generic template.
- LLM prompts, rule engines, and data flows tuned to your criteria and risk lens.
- Outputs designed to drop into your existing models, memos, and pitch materials.
Once the workflow proves itself on real deals, we harden it, deploy it into your environment, and iterate as your team uses it.
- Clear documentation for analysts and associates.
- Light-touch support or deeper ongoing collaboration, depending on need.
AI Engineering for Private Markets
I’m Michael DeAngelis, and I sit at the intersection of leveraged finance and modern AI systems engineering.
As an Analyst at Sigiriya Growth Advisors, I work on private credit and capital advisory mandates while leading the build-out of internal AI tooling — including an LLM-powered underwriting engine, deal intelligence workflows, counterparty matching logic, and internal data systems used by the team on live mandates.
Before that, I worked in Cloud Support Engineering at Amazon Web Services (AWS), delivering serverless data pipelines with S3, Lambda, EventBridge, Step Functions, and DynamoDB. That mix of deal experience and AWS-native AI tools (Bedrock, SageMaker) is what Marrett Advisory brings to your firm: innovative AI systems that are financially literate, technically sound, and shaped around how your team already runs deals.
- AI-first, not AI-only: The point isn’t to bolt “AI” onto everything. It’s to quietly automate the 30–70% of each workflow that’s repeatable, and let your people handle the judgment.
- Flexible by design: Start with one workflow (screening, counterparty matching, CIM drafting, IC memos, etc.), adjust based on live use, and expand only if it clearly helps.
- Innovative but grounded: I use modern AI (LLMs, RAG, embeddings, serverless architectures) in ways that fit existing deal processes, compliance constraints, and real-world time pressure.
Start with one AI-powered workflow.
Think about where your team spends the most manual time: screening CIMs, mapping deals to the right counterparties, drafting CIMs or pitch books, preparing IC materials, or tracking portfolio information. There is almost always one process that’s obviously painful and clearly repeatable.
Send me a short description of that workflow and a couple of anonymized examples. I’ll outline how AI and modern AWS infrastructure can automate it securely inside your environment. We’ll scope the work to your firm’s size and budget, and only expand if the value is clear on real deals.