Run one repeatable loop from grant approval to the LP report: book the onboarding call, lock the impact agreement, and let Sopact Sense chase missing data and flag variance automatically.
Run one repeatable loop for every grant or investment: an onboarding call, an impact agreement, a tracking cadence, and an LP or grantee report — with Sopact Sense carrying the agreement forward automatically.
In short: the moment a grant or investment is approved, start with whichever input you have — a 90-minute onboarding call transcript, notes from the call, or a theory of change the grantee or investee already has. Bring it into Sopact Sense. Sense builds and audits the framework, drafts the follow-ups for missing data, tracks the agreed cadence, and flags anything unusual before it reaches your LPs. The same loop runs for one investee or a hundred.
Most portfolio and grant managers rebuild this cycle from scratch every time — a spreadsheet here, an email thread there, a reminder that lives in someone's calendar. The fix isn't a bigger spreadsheet. It's making the onboarding call the one place the cycle starts, and letting Sense carry the agreement from there.
The eight steps below walk the full cycle, with copy-paste prompts for each stage.
Set the 90-minute onboarding call within days of approval — not whenever calendars align weeks later. This is the one conversation where the portfolio manager and the investee agree on what "done" looks like: which framework, which metrics, which documents.
You don't need a clean intake form. Sense accepts the raw call transcript, the investee's public program page, or a framework file they already use (Theory of Change, Logic Model, Logframe, Results Framework). Point it at whichever exists:
You are the Sopact Sense Assistant. Here is my onboarding call transcript / program page / existing framework for [INVESTEE OR GRANTEE NAME]: [PASTE TRANSCRIPT, URL, OR FRAMEWORK]. Use only what it states; mark anything you infer as [INFERRED]. Wait for my task.
One prompt does two jobs — it converts the intake into the investee's chosen framework, then audits it the same way Sense audits any framework: naming the assumption under every link and grading each element by evidence.
Convert this intake into a [THEORY OF CHANGE / LOGIC MODEL / LOGFRAME / RESULTS FRAMEWORK] for [INVESTEE]. Then audit every element: name the assumption under each link, flag what is missing, vague, or unevidenced, and grade every element green / amber / red (green = specific + evidenced, amber = vague or unevidenced, red = missing). Use only what the intake states; mark anything inferred [INFERRED].
On the demo portfolio (Meridian Impact Fund, DEMO-06 · 12 investees), the spread is real: high-evidence investees clear mostly Green, mid-quality investees land a Green/Amber mix, and thin-data investees — beneficiary counts with no outcome measure — grade mostly Red.
GRADE: green | Verified outcome, evidence quality 4/5 | Bright Path Education — baseline + endline on persistent IDs; amber | Financial documents pending | Riverside Youth Collective — budget referenced on the call, not yet uploaded; red | No outcome metric | Riverside Youth Collective — beneficiary counts only, evidence quality 2/5
In addition consider on the following validation and alignment as per your impact measurement and management strategy
Every amber or red element turns into a specific, named ask — not a generic "please send your data" message. The portfolio manager reviews and edits before it goes out.
From the amber and red elements above, draft a short email to [INVESTEE CONTACT] asking for the specific missing pieces: [KEY OUTPUT/OUTCOME METRICS / FINANCIAL DOCUMENTS / SOCIAL AUDIT]. Name exactly what's missing per item and why it's needed — no generic "please send your data" language. I will review and edit before sending.
This loop repeats weekly, automatically, until the impact agreement's three required components are all in: key output and outcome metrics (qualitative and quantitative), financial documents, and a social audit or equivalent.
Once the three components close, the agreement is locked and the reporting cadence — quarterly, semi-annual, or annual — is set from it directly. Sense drafts the recurring data-request message for that schedule so no investee is tracked on a cadence nobody remembers agreeing to.
The impact agreement for [INVESTEE] is now complete. Set a [QUARTERLY / SEMI-ANNUAL / ANNUAL] reporting cadence and draft the recurring data-request message for that schedule, referencing the exact metrics and documents agreed to above.
When data comes in — documents, open-ended responses, and structured metrics together — Sense rebuilds the framework-aligned report from that submission and re-grades every element against the original audit. No separate review pass for the narrative, the numbers, and the attachments.
[INVESTEE] has submitted this cycle's data: [PASTE DOCUMENTS, OPEN-ENDED RESPONSES, AND METRICS]. Rebuild the framework-aligned report using only what was submitted; mark anything inferred [INFERRED]. Re-grade every element against the original audit.
Before anything reaches an LP, Sense compares the new report against the impact agreement's targets and the prior cycle, and produces a short review queue — not a re-read of the whole submission.
Compare this cycle's report to the impact agreement's targets and the prior cycle. Flag: (1) any unusual insight, (2) any field still missing or vague, (3) any metric with material variance from target. For each flag, state the number, the evidence, and what needs correcting.
The portfolio manager works the queue directly: approve what's clean, and send the rest back with a specific, named ask — same discipline as step 4.
For the flagged items marked "needs correction," draft a short message to [INVESTEE CONTACT] naming the specific gap or variance and what would resolve it. I will review before sending.
With the queue clear, roll the cycle up into a decision-first report for the LP committee, in your own brand, and create a shareable link.
Create an LP-ready report rolling up [INVESTEE OR "the full portfolio"] against the impact agreement, in Sopact branding [or paste your fund's brand guideline]. Lead with the decision this report informs. Then create a shareable link and open it in a new tab.
Start the clock at approval, not at "when we get to it." The 90-minute call loses most of its value if it happens a month after the grant clears — the team's own memory of what they promised has already faded.
Never send the generic reminder. A follow-up that names the exact missing metric gets answered in days; "please send your Q3 data" gets answered in weeks, if at all. Let Sense draft the specific ask every time.
Treat the review queue as triage, not re-reading. The queue exists so the portfolio manager reads three flagged lines instead of the entire submission. If you're re-reading everything anyway, the flags aren't specific enough — ask Sense to restate each one as a number plus what would resolve it.
Tighten the impact agreement template while you're here. Once a cycle or two has run, ask Sense which fields are chronically vague across the portfolio — that's usually a sign the agreement's own wording needs to change, not the investees.
Across the last two reporting cycles, which fields come back vague or incomplete most often, and for how many investees? Suggest a rewording of the impact agreement template for those fields so future responses are specific by default.
An impact agreement is the set of measurement commitments a funder and an investee or grantee lock in at onboarding — which framework applies, which key output and outcome metrics will be tracked, and which financial documents and audits (like a social audit) are required. It's the reference point every later report and reminder is checked against.
Set one reporting cadence per investee at onboarding — quarterly, semi-annual, or annual — tied to their impact agreement, and let an AI assistant like Sopact Sense draft the recurring data requests, rebuild the framework-aligned report from each submission, and flag missing or vague fields automatically. Consistency comes from the cadence and the agreement being fixed once, not re-negotiated every cycle.
Name the specific gap — which metric, which document, which field — rather than sending a generic reminder. Draft the follow-up automatically from the framework audit's amber and red elements, review and edit it, then send. Repeat weekly until the impact agreement's required components are complete; a specific ask closes faster than a generic one.
ChatGPT, Claude, and Gemini are fine for a quick test — but not for an answer you'll put in front of a funder or board. When it has to hold up, run it in Sopact Sense.
Try it in Portfolio Intelligence →
