Sopact is a technology based social enterprise committed to helping organizations measure impact by directly involving their stakeholders.
Copyright 2015-2026 © sopact. All rights reserved.
One client record, read on arrival — from intake to outcome.
Case management software is a platform that holds one persistent record per client across the whole service lifecycle — referral, intake, assessment, service or treatment plan, case notes, services delivered, and outcome follow-up — so caseworkers, supervisors, and program directors get answers without stitching spreadsheets together. It is also called a case management system, client management system, or client tracking software. The newest generation adds intelligence to the record itself: AI reads each case note on arrival, scores it against the service plan with citations, and turns follow-up surveys into outcome evidence — so the software answers whether the client's situation actually improved, not just what services were logged.
Used by: nonprofit human services · social work and family services · workforce and reentry programs · housing and homelessness services · community action agencies · behavioral-health-adjacent and public-health programs · multi-program agencies coordinating one client across teams.
Not because the software stopped working — because storing the client and logging the service became table stakes. Bonterra (Apricot, ETO, Social Solutions), Salesforce, Penelope, Casebook, ClientTrack, and CaseWorthy earned their positions honestly: they got casework out of the filing cabinet, standardized the intake, and gave agencies a real system of record. If your problem was collection — hundreds of clients, a dozen caseworkers, one funder deadline — that generation solved it.
But the strengths hardened into weaknesses. Implementations run three months to a year; agencies describe legacy builds where every new program or funder change is another integrator engagement. The logic is rigid by design, and the richest evidence those systems hold — the case notes — dies after collection: narrative piles up in a system nobody reads across, and the analytics describe what was delivered, not what changed. A program lead running services across sites put the pattern plainly: they could track "outputs and dollars and volunteer hours," but "the biggest pain point is that we can't tell a cohesive story across all of it. Each thing works fine in a silo."
The cost of not reading is concrete. As one practitioner described it, case notes end up "just sitting around in the systems… by the time they find out, you already failed a child." The work that decides whether a program is well run has moved to the two ends the record-keeping tools never owned: reading every case note and assessment on arrival, and proving the outcome months and years later on the same client.
None of this requires ripping out your incumbent. The sentence we hear on almost every call now: "We're not gonna leave our system, but we're open to an AND." Keep the system of record; add the layer that reads what it produces. (If you're comparing named platforms outright — rankings, pricing, reviews — start with best case management software.)
The stake, stated honestly: boards and funders have already changed the question from "how many did you serve" to "did their situation improve, and can you show it." If you are signing a multi-year configuration build today, ask which question it will be able to answer when it finally goes live.
Case intelligence is reliable answers from your case data — in minutes, not months. Everything a client touches is treated as data: the referral, the intake assessment, the validated screen, the service plan, every case note, the 90-day and year-three follow-up. All of it lands on one persistent client record, aligned to your framework and data dictionary, so the same person looks like the same person across three programs and five years.
The part that changes daily work is the Assistant. Caseload analysis, screen scoring, and open-text case-note analysis are unified into one chat-based function: ask a question, get a defensible answer with citations to the underlying records. No prompt engineering, no dashboard hunting, no waiting for the one analyst who knows where the export lives. A program is never one user — caseworkers, supervisors, finance, the board, funders, and the clients themselves all need different views of the same record — and a chat interface empowers each of them directly.
When the analysis is done, it does not die in the chat: create shareable reports tailored to each audience — the supervisor's caseload view, the funder outcome report, the board summary — from the same underlying answer, each number traceable to the source case note.
One proof point from the field. Open Play Foundation ran programs the way most funded organizations do — intake forms, follow-up reflections, stacks of narrative that never made it past the spreadsheet. When that work moved onto Sopact, the record could finally read itself: "Those statistics that we're now running on Sopact immediately showed me there's something significantly wrong … things like that, we would never have been able to do in the past." — Marco Botha, CEO, Open Play Foundation. A system of record tells you the service was delivered. An intelligent record tells you something is wrong in time to act.
The honest way to evaluate case management software is against the lifecycle, not the feature list. Nearly every framework describes the same arc — and it is a loop, not a line: monitoring routinely sends a case back into reassessment until the goals are met. Below is the full cycle — six stages, each with what the software should do, the exact prompt to use, and what to expect back. Every prompt is copy-paste; the placeholders in brackets are yours to fill.
Intake is where clean-at-source pays or fails. Instead of free-text answers you will pay a caseworker to decode later, the form is designed so every narrative field maps to your framework, and every client gets a persistent unique ID that follows them across programs and years. Eligibility screening, consent capture, save-and-return — and AI drafts the intake form from the program documents you already have.
Build a client intake form from this program description: [PROGRAM URL OR DOCUMENT]. Create structured fields for demographics, eligibility, and consent; narrative fields for presenting needs and goals mapped to our theory of change; and eligibility screening questions with clear pass/fail criteria. Flag any question that collects information we already hold on returning clients.
Expected output. A ready-to-edit intake form: structured fields, mapped narrative prompts, eligibility gates, and a persistent client ID assigned at first contact.
Tips for reliable output. Give the AI your theory of change and data dictionary before form design. Assign the client ID at referral, not at enrollment — everything downstream attaches to the ID created here.
The baseline is the reference every later wave is compared against. Validated screens (PHQ-9, GAD-7, VI-SPDAT, or your own indicator) and the intake narrative land on the same record, and the assessment is read the moment it arrives — needs, risk factors, and protective factors extracted and cited, not left in a folder until something goes wrong.
From this intake assessment, extract the client's baseline needs, risk factors, and protective factors, each with the exact source sentence. Score the validated screens included, flag any safeguarding or immediate-risk language for human review, and note where the assessment is incomplete. Do not infer a diagnosis — report only what the text supports.
Expected output. A structured baseline with per-item evidence, scored screens, and a flagged list of risk or safeguarding language routed to a human.
Tips for reliable output. Lock the baseline before services begin — a baseline captured on day one, even on a handful of clients, proves the loop works before anything scales.
Every client gets a service or treatment plan built from the assessment and mapped to your framework — goals that are observable, timelines that are real, and the outcome each service is meant to move. The plan becomes the thing case notes are later read against.
Draft a service plan from this assessment: [ASSESSMENT]. Map each identified need to a measurable goal, a service or referral, and the outcome indicator it should move, aligned to our theory of change. Write goals as observable statements a caseworker can evidence, and flag any need with no service currently available.
Expected output. A service plan with measurable goals, mapped services, outcome indicators, and a gap list where needs have no matching service.
Tips for reliable output. Name the outcome for every goal. A plan that can't say what success looks like can't be evaluated later.
This is the stage record-keeping software cannot do. Every case note is read as it lands, coded against the service plan, with risk signals — missed appointments, disengagement, safeguarding language, escalation — surfaced the week they appear instead of at the supervisor's month-end sample. The narrative stays with the caseworker; the structure is generated and tied back to the source sentence.
Read this batch of case notes: [NOTE BATCH]. For each client, summarize progress against the service plan with citations, code the note against our outcome indicators, and flag risk signals — missed appointments, disengagement, safeguarding or escalation language — with the exact source sentence. Use the same method as last month so results are comparable.
Expected output. Per-client progress summaries with citations, coded outcome evidence, and a risk-flag list with sources — the day notes are written, not six weeks later.
Tips for reliable output. Route every risk flag to a named owner with a deadline. A flag nobody owns is a finding that sat there.
Closure is not the end of the record. The 90-day, one-year, and three-year follow-ups land on the same client ID as the intake assessment, so the question every funder asks — did the situation actually improve — has a reproducible answer instead of a year-end reconstruction. Re-engaging clients arrive with their full history attached.
Compare the baseline assessment to the [90-day / 1-year] follow-up across [COHORT / PROGRAM]: which outcomes moved, by how much, and with what confidence? Show change per indicator, note where the sample is too small to conclude, and pair every number with a representative case-note quote. Treat this as change over time, not attribution.
Expected output. A baseline-to-follow-up outcome analysis with honest confidence bounds and a narrative quote behind each number — the longitudinal view a persistent ID makes possible.
Tips for reliable output. Capture contact channels and follow-up expectations at intake, not at exit. The longitudinal horizon is what separates an exit survey from an outcome.
Reports are questions, not formats. From the same accumulating client record, the caseload report, the supervisor dashboard, the funder outcome report, and the HMIS or CSBG ROMA submission are each one query — with the supporting case note two clicks away — instead of a two-to-four-week reassembly across intake, services, and follow-up systems.
Aggregate this program's client records into a [funder] outcome report: outcomes achieved against targets, coded case-note themes ranked by frequency with representative quotes, demographic distribution, and clients flagged as missing a required follow-up. Cite the source client record for every number and quote. Format one version for the board and one for the funder.
Expected output. A funder-ready outcome report generated as a query, every figure citing its source record — plus the "missing" list surfaced before the deadline asks.
Tips for reliable output. Lock the data dictionary before the first reporting cycle and version every change — comparability across years is the entire value. If your outcome framework needs an external anchor, align it to IRIS+ so metrics are comparable beyond your own walls.
Beyond table stakes — intake, caseload views, services logging, security — four criteria actually separate tools: time to first live cycle (days vs. a quarter), whether AI reads case notes on arrival or a supervisor still samples them by hand, whether configuration is plain-English or a consultant engagement, and whether the platform can prove client outcomes rather than just count services. Ask every vendor to show the outcome report on real data, not a slide.
The evaluation itself is work you can delegate to AI. These prompts mirror what buyers are already asking answer engines — use them as they are:
Build an evaluation matrix for case management software with technical and program criteria weighted 50/50. Technical: security and field-level access control, integrations with our HMIS or billing system, configuration model, data export and exit rights. Program: AI case-note reading with citations, one client ID across programs, longitudinal outcome tracking, funder report generation. Score vendors [VENDOR LIST] on each criterion with evidence required, not vendor claims.
Propose a 30-day pilot plan to evaluate case management software: one program, one cohort of roughly 50 current clients, tested end to end (intake → assessment → service plan → case-note reading → outcome report), with numeric success thresholds and rollback criteria if the pilot fails.
A note on scope while you evaluate: the same spine adapts by vertical with a different intake and funder report — nonprofit case management software, social work case management software, human services case management software, and housing case management software each cover their fit directly. For the case note layer itself, see case notes software; for the front door, client intake software.
The stages above are the argument; the Academy articles are the practice — each a hands-on companion for one workflow, written to run on your own data.
Honest boundaries, because the fastest way to a failed implementation is buying the wrong category.
Not a CRM, and not an EHR. A CRM (Salesforce, Blackbaud) tracks donors and relationships; an EHR (Epic, Cerner) tracks clinical encounters and billing. Case management software tracks clients through a service-delivery lifecycle, with case-note narrative and outcome evidence at the center. A "case management CRM" is a CRM stretched to do casework — see case management CRM for where that fits and where it doesn't.
Not your billing or HMIS system. The general ledger, the payment engine, and the HMIS submission stay in the systems built for them; case intelligence integrates on one shared record rather than replacing them.
Not for every compliance regime. Sopact provides AES-256 encryption, TLS 1.3, field-level role-based access, SSO/MFA, and full audit logging, with AI under enterprise SLAs and no training-data retention — but Sopact is not currently HIPAA-certified or covered by a Business Associate Agreement. If your program is subject to HIPAA, FERPA, 42 CFR Part 2, or county behavioral-health rules, evaluate these controls against your compliance program and confirm scope in writing before storing protected information. And if your use case is purely a data warehouse, Sopact is not the ideal system for that.
Case management software is a platform that holds one persistent record per client across the service lifecycle — referral, intake, assessment, service plan, case notes, services delivered, and outcome follow-up — so caseworkers, supervisors, and program directors get answers without a spreadsheet merge. Also called a case management system or client tracking software. The newest generation adds AI that reads each case note on arrival and turns follow-up into cited outcome evidence.
The best case management software for outcome tracking is one that reads every case note on arrival rather than waiting for a year-end review, keeps one persistent client record across every program a client touches, generates outcome reports as queries instead of CSV merges, and supports longitudinal follow-up at 90 days, one year, and three years on the same client ID. Legacy choices like Bonterra Apricot, Salesforce, Penelope, and Casebook were built for a configure-heavy era; AI-native platforms like Sopact treat configuration as conversation and generate outcome reports as queries.
Sopact is priced by use-case complexity, not by seats or caseload: how many programs share the client record, how custom the data dictionary is, which built-in skills are activated, longitudinal depth, white-label depth, and API integration to HMIS or billing. A 12-person family services agency running one program pays less than a 50-person multi-site community-action agency running six. Free and spreadsheet options cover basic service logging, but the cost moves elsewhere — lost continuity when staff turn over, manual outcome reporting, and the staff hours spent reconciling intake, services, and follow-up at year-end.
Yes — when it is built around one persistent client record. A family receiving social work services, workforce navigation, and housing support appears as one record with three service streams, not three separate cases. Multi-program agencies use this to coordinate across teams, recognize re-engaging clients automatically, and report at the agency level instead of reconstructing the picture per program at year-end.
AI reads every case note on arrival and codes it against the service plan and the outcome screens the client completed at intake. Risk signals — missed appointments, disengagement, safeguarding language, treatment-plan deviations — surface the week they appear instead of at the supervisor's monthly review. The narrative stays with the caseworker; the structure (themes, indicators, attribution to outcomes) is generated automatically and tied back to the source sentence. For the note layer specifically, see case notes software.
Look for AES-256 encryption at rest, TLS 1.3 in transit, role-based access to the field level, SSO with MFA, and full audit logging — all of which Sopact provides, with no training-data retention on AI calls. Sopact is not currently HIPAA-certified or covered by a Business Associate Agreement; if your casework touches protected health information under HIPAA, FERPA, or 42 CFR Part 2, treat that as gating and confirm scope in writing. Sensitive fields can be excluded from AI processing entirely, and analysis can run on anonymized IDs.
Both were built for the configure-heavy era. Bonterra Apricot is closer to community-services casework out of the box with prebuilt intake and outcome templates; integrator work is still required for outcome reporting. Salesforce (Nonprofit Cloud, or Public Sector for government) is more flexible but requires significant integrator work to become a real case management system, and Einstein AI is an add-on rather than native. Both produce numbers; neither reads case notes on arrival. AI-native case intelligence avoids both the configuration cost and the year-end reporting cost — and often runs alongside an incumbent as the reading layer rather than replacing it.
Six questions separate platforms that work from platforms that only demo well: How long until our first cycle is live? Will every case note get read, or just the ones a supervisor samples? Does the same client appear as one record across all our programs? Is the caseload report a query, or a CSV merge? When a funder or auditor asks why this outcome, can I show the supporting case note in two clicks? And does the system work offline for field work? Then run a contained pilot on your own data before you commit.
Two months, one contained use case — one program, one intake form, one cohort of clients you already serve. You bring last year's case notes; the pilot shows you the coded, cited version of your own caseload, ending with a demonstrated export. If the outcome answers aren't defensible in front of your board or funder, don't continue. Scope a 2-month pilot →