A Private Working Group Built Around
a Shared Agenda.
Participation is invitation-only. The Circle brings together managing partners at impact funds who are facing the same kind of challenge, and who actually want to sit with it and work it through as a group.
Circle members convene around defined programs, each intended to tackle a clear operational need. The program looks at AI across the entire investment lifecycle, in eleven stages, from formation and origination through screening, diligence and the investment committee, then into value creation, portfolio monitoring, LP reporting, exit and wind-down.
Five Funds Finished With Working Systems,
Not Notes.
The first cohort ran from July to September 2026 with five impact funds across Africa, Asia, Latin America and Europe, investing in equity, debt and grants. Each fund built inside its own environment on its own documents. These are their results, described without naming the funds.
A Latin American debt manager wrote an AI policy aligned to its own recently updated work processes across investment, ESG, technical assistance, asset liability management and treasury, rather than adopting a generic template. He then built a skill that reads the firm’s AI usage data and reports quarterly on how each team and each person is using it.
A European venture investment fund built an internal assistant on its own document corpus, added organization-wide skills and a data map of its document store, and set out four governing principles for AI use across the organization. The fund presented the whole build to the cohort.
A financial inclusion equity fund now runs a shared project per company under diligence, so the deal team works from one place rather than from individual chat histories. Financial modeling on live deals is materially faster.
A North African growth equity manager arrived a year into daily AI use, with prompt libraries, skills and valuation agents already running. The addition it named as the biggest was workflow orchestration, the layer that connects one step to the next without a person in between.
One member is now running his own internal sessions with two colleagues on the origination team, working through the same material in his own language and his own deal context. That is the adoption test the program is built to pass.
Two members at opposite ends of the maturity range said the same thing in the closing session, four minutes apart and without hearing each other: the work comes before the relief. Time has to be carved out before any of this gives time back. We put that in the offer rather than in the small print.
I joined the AI Circle hoping to build a broader understanding of AI. What surprised me most was not only the capability of the technology, but how accessible it already is and the potential it offers to rapidly scale social impact in ways that were previously difficult to imagine.
The biggest shift for us was going from asking how each person can use AI to thinking about how the fund itself can work differently with AI.
It helped us go to the next level of AI usage. It helped us automate part of the investment process.
Exploration of the use cases for AI in private equity has been mind boggling. It goes from pipeline management to portfolio management, deal sourcing to due diligence, value creation plan to LP reporting.
Every Session Ends
With a Working Asset.
The Circle is a working group. Over seven weeks we move through the eleven stages of the investment lifecycle in the order a fund actually meets the work, starting with what your team does every week. Every session opens with a problem you are probably already sitting with, and ends with something your team can use the next day.
Level 1 is a prompt run by a person. Level 2 is a workflow that chains several steps together and reads a store of record, so the output is the same whoever runs it. Level 3 is a system that runs without being asked. Most funds arrive somewhere in Level 1. Cohort 1 finished asking for Level 2, so Level 2 now sits at the front of the program rather than at the end.
Large Enough to Need a System.
Small Enough to Move Quickly.
The Circle works best where real AI use is already happening but has not yet been shaped into a fund-level operating system. That is the gap we close.
A decision process
You run an investment committee, or an equivalent body that decides on individual transactions. Equity, debt and grantmaking all qualify on that test.
LP accountable
Your ESG and responsible investment commitments mean how you use AI matters to your LPs, well beyond a policy document.
Real usage already
Your team uses AI informally. The Circle turns that into a shared fund-level system with a named owner per stage.
If your team has no AI exposure at all, start with the AI Intensive below. If you cannot protect 90 minutes a week for seven weeks, the cohort will not work for you.
Most AI programs focus on fund operations. We also help you ask better questions about AI use inside your portfolio companies, particularly where customer protection, regulated financial data, or inclusion outcomes are involved. For each session where a portfolio company uses AI in a customer-facing process, we provide a structured set of board-level diligence and governance questions.
Seven Sessions, Ordered by How Often You Do the Work.
The eleven stages of the lifecycle are the map and they never move. The three levels are the ladder. The running order comes from frequency and pain, so the program starts with the recurring weekly load and closes on the team standard. Click the + next to any session to expand.
More teams are stopped by an administrator toggle than by anything in the curriculum. Skills are switched off at the workspace level, code execution is disabled, personal accounts sit alongside enterprise ones, and nobody has written down which categories of fund data are allowed to go where. In cohort 1 a live session lost sixteen minutes to account permissions before a single piece of content was delivered.
Tool access cleared across the seats you are sending, with the specific administrator settings named. A data classification across three tiers, covering what may go into a general model, what stays inside your tenant, and what never leaves your systems, with the jurisdiction principle applied to your markets. A readiness score per component, which becomes your baseline.
Session zero is also open on its own, free, before you join. See Set Your Baseline →
Teams describe their AI use case by case, one person at a time. Nobody can see which parts of the fund are covered, which are untouched, and which have three people solving the same problem separately. Without a map, every conversation about AI becomes a conversation about tools.
Your own lifecycle map across all eleven stages, marked with where your fund sits at each one, as a working document rather than a slide. Your Fund Thesis System Prompt, the foundation that makes every later tool thesis-aware, including a responsible AI lens set at configuration.
A good associate spends four hours on a CIM before the managing partner says no in three minutes. In frontier markets the problem compounds: information memoranda rarely follow a standard format, financials sit inside narrative PDFs, and comparable transaction data for a Series A fintech in West Africa is thin at best. There is no sector database for early-stage digital financial services in Bangladesh or climate-smart agribusiness in Kenya, so every new sector search starts from scratch and most of that work falls to the partner.
A screening skill running on your own pipeline, not on a worked example. Built as a ladder: a four-step quick screen, the full screen, and the same screen loaded with your own decision history so a reopened deal is not screened as though it were new. A Sector Monitoring Blueprint that works with Tracxn or CB Insights where you have them and without them where you do not. A Decision Log Backfill, which reconstructs your past investment decisions into a structured record the screen can read.
You are on the board of seven companies across four time zones. Management accounts land 48 hours before the meeting. Your DFI co-investors expect evidence of active board engagement rather than board minutes. You arrive having read everything and thought about nothing, because there was no time to think. Stress signals in frontier market portfolio companies surface later than they should, and by the time the quarterly report flags a currency move or a regulatory change, you are three months behind.
A value creation workflow chained across one of your portfolio companies, which is the first Level 2 asset most funds own. A Board Briefing Workflow covering performance extraction, open action tracking, strategic question generation and partner authentication. A Portfolio Monitoring Exception Report sized for a twenty-minute monthly review, with five responsible AI signals added: customer protection, exclusion risk, bias risk, model governance and impact integrity.
Capital raising takes 30 to 40 percent of partner time with almost no AI in the workflow. At the other end of the same relationship, impact reporting for DFI-backed funds means IRIS+ metrics, SDG mapping and separate cross-walks for FMO, IFC, BII, OeEB and AfDB, each in a different format. The data already exists. Assembling it costs three days of partner time every quarter. Formation and reporting read the same corpus and speak to the same audience, so they belong in one session.
An obligations register built from your own side letters and LP agreements, turning commitments buried in legal text into structured fields your team can track. One generated report produced end to end from your own data. An Impact Measurement Mapping Prompt for the IRIS+ to SDG to DFI framework cross-walk, and an LP Narrative Drafting Workflow.
A deal-killer found after 150,000 dollars in legal fees was avoidable, and that question had an answer three months earlier. Diligence in frontier markets has no standard playbook: regulatory filings may be incomplete, local advisors work on different timelines, and a market visit that would resolve a key question costs two days of travel. The five questions whose negative answer should stop the deal often go unasked until the legal engagement is already running.
The diligence prompt ladder run against a real data room, yours or a supplied one. Five Kill-Shot Questions defined before legal engagement begins, with responsible AI diligence questions for high-risk portfolio companies. A Three-Step Verification Protocol and an AI governance checklist. A Local Advisor Call Protocol for preparing, capturing and synthesising expert calls, and for surfacing contradictions between what advisors say and what the documents show.
IC memos look different deal to deal. When the managing partner approves a deal, the reasoning stays in their head and is never written down in a form the team can learn from. Pattern recognition across twelve investments is out of reach when each one is documented differently. This is also the point in the lifecycle where AI stops summarising and starts acting, and where third parties begin holding your data, which is why governance and compliance markers sit in this session rather than in a policy appendix.
A committee pack draft generated from your own diligence output, with a responsible AI section standard in every memo. A conditions precedent tracker for the closing period. A Term Sheet Benchmarking Prompt for pre-negotiation intelligence, built on a term library assembled from your own past deals. A Decision Record Protocol, which turns each approval into institutional memory the screening skill in Session 2 can read.
This is the finding cohort 1 produced most clearly. Every individual skill worked. No fund turned one into a team standard without somebody convening it. AI arrives at enterprise level, each person builds their own version of the same thing, and the fund ends up with five private ways of screening a deal and no agreed one. Governance is usually missing at the same time, particularly at smaller standalone funds.
An AI policy written against your own work processes rather than adapted from a generic template. A named owner per stage, so each part of the lifecycle has somebody accountable for the standard. A usage and evaluation approach, because a practice that is not measured does not improve. The handover pack: your complete Fund AI operating system across the eleven stages, with every prompt, skill and workflow indexed and linked.
Exit and wind-down ship as written self-serve packs rather than live sessions. Funds that want to go further after Session 7 can move into a portfolio company AI program with Accendo, or into a build engagement on two or three stages at Level 2. Both are separate engagements, not part of the cohort fee.
Set Your Baseline.
A Free Hour, Before Either Path.
Session zero, the same one every Circle fund starts on, is also open on its own and at no cost. No invitation required. You leave with tool access sorted, the interface basics, one thing you built yourself, and a clear view of which path fits your fund, the Circle or the AI Intensive.
The AI Intensive:
The Whole Lifecycle in Five Hours.
Not everyone is ready to commit a fund to seven weeks, and not every seat needs to. The AI Intensive is the end-to-end walkthrough on its own, open to individuals, with no invitation required. It is also the shortest way to find out whether the Circle is right for your fund.
Courses Teach You About AI.
The Circle Builds Your Fund’s System With You.
You can take a course online, attend a well-regarded university program, and watch a great deal of video. Each of those leaves you with the same task: apply the lesson to your own fund, your own thesis, your own documents. That is the part that does not happen, because the partners who most need it are busy with origination, boards, exits and fundraising.
The Circle is curated. We meet you beforehand to work through your AI readiness and match you with peers who are facing comparable problems and are not competing with you for deals. Every session then builds one asset inside your environment, on your material.
What you keep is concrete: skills your team can run, workflow templates you can change, an indexed library of every prompt and tool arranged by lifecycle stage, and a short list of invest-tech platforms worth your attention. The program ends. The system stays with the fund.
Common Questions, Answered Directly.
The AI Intensive is charged per seat. A Circle place is priced per fund, in bands set by assets under management, or by annual capital deployed for grantmakers, and covers up to four seats. We send the rate card with the readiness questionnaire so the number arrives alongside the curriculum rather than ahead of it. Fees are invoiced in US dollars and settled before the first session.
Yes. Session zero, the same session every Circle fund starts on, runs as a free one-hour session called Set Your Baseline, open to anyone, no invitation required. It runs Mondays and Fridays at 07:00 New York, ten places a session, and you can register here. You leave with tool access sorted, the interface basics, one thing you built yourself, and an overview of the Circle and the AI Intensive, so you can judge which path fits before committing to either.
Ninety minutes a week for seven weeks, plus office hours you use as you need them, plus the time your team spends between sessions. Two members of cohort 1, one at the start of their AI journey and one well into it, made the same point in the closing session: the work comes before the relief. Expect the first three sessions to cost you time and the value to land from session four onward. Funds that cannot protect the weekly slot should take the Intensive instead.
Yes, and we settle it before you join rather than during. The frameworks are platform-independent. The prompts and skills are built first for Claude, and we ship the Copilot equivalent for the ones your team will actually run, including the Agent Builder path where an instruction set is too long for the field. Cohort 1 had funds on both, and the platform position is now confirmed at the invitation stage so nobody discovers the gap mid-program.
Cohort 2 is equity. A debt cohort runs from January 2027 and we are taking names for it now. The fit test is the decision process rather than the instrument: if you run an investment committee or an equivalent body deciding on individual transactions, the material applies. Grantmakers qualify on the same test and are priced on annual capital deployed rather than assets under management.
Yes. The curriculum follows the eleven stages of the investment lifecycle, and those stages are the same whether you invest in fintech, agritech, climate infrastructure or financial services. Where an in-session exercise needs a deal, you can use your own or a supplied mock pack, so worked examples stay concrete without putting your live deal data in front of the room.
That is normal, and session zero exists to level it. The Circle is designed for a managing partner and their immediate deal team to build a shared way of working. The implementation seat, a COO, chief of staff or platform lead, is specifically included to drive adoption after the sessions end, because the single clearest finding from cohort 1 is that the unit of adoption is the team rather than the individual.
Between sessions there are open office hours for application questions, and each participant posts one short update to the cohort group within 72 hours of each session: one thing tried, one result, one question. After session seven you keep the full indexed library and the handover pack, and there is a 90-day check-in. Funds wanting deeper support can move into a separate build engagement with Accendo.
The Circle is a program of Accendo Associates LLC. Accendo works with impact investors, development finance institutions and financial service providers on AI and digital transformation, and designs and delivers every session of the Circle directly.
The Readiness Questionnaire
Is the Next Step.
Before we confirm any place, we send a short readiness questionnaire. It asks where your fund uses AI today, your jurisdiction and investment focus, which platform your team runs on, and whether the timing works for you. It also unlocks the full curriculum and the rate card. Enrollment for cohort 2 closes on 8 October 2026 and the first session is on 15 October 2026.
Request the readiness questionnaire →