CivicLoop by Ta-Tech Solutions Purpose: Know exactly who we are up against, where every one of them fails, and the specific lane CivicLoop takes to win the Prince George's County 311 / Community Relations engagement.
Prince George's County currently runs PGC311 - a web portal
(pgc311.com) plus iOS and Android apps, operated by the Office of
Community Relations, built by a vendor called Connected Bits
(the Android package is literally com.connectedbits.connectedcounty311).
It is failing, measurably, right now:
This changes our entire approach. We do not have to argue in the abstract that a better 311 is possible. We show PGC311 failing to submit a request, then submit the same request through CivicLoop in under a minute, by voice, in Spanish. A live, demonstrable incumbent failure is the rarest opening in government procurement. We build the whole pitch around it.
Prince George's County also runs MyPGC - "sign up for emails, alerts, notifications" - a separate opt-in resident-communication system, distinct sign-up, distinct vendor (the "emails, alerts, notifications" pattern is the signature of Granicus govDelivery, the outbound-comms market leader).
So the County's resident-facing service layer is split in two:
A resident must know about, and sign up for, both - and even then, the inbound system does not talk to the outbound one, so reporting a pothole does not subscribe you to news about that pothole. This is the fragmentation failure mode (Section 3) made concrete. CivicLoop's living record collapses inbound and outbound into one thing: report it, and you are automatically and specifically informed about it. CivicLoop can also feed MyPGC rather than replace it - the anti-lock-in posture in practice.
The 311 / citizen-services software market is real and crowded, but the leaders lead on distribution and procurement relationships, not product quality. That is the structural weakness we exploit.
SeeClickFix (owned by CivicPlus) - the category-defining resident-reporting app. Web + mobile, photo/GPS reports, public map, status updates. Widest small-and-mid-market footprint, strong resident brand recognition. But it is a reporting funnel, not an intelligent system: weak analytics and reporting, no integrated work-order or labor tracking, poor integration with other municipal systems (duplicate data entry), and effectively no AI. CivicPlus sells it inside a broad bundle (websites, agendas, notifications) where each module is middling.
Accela - owns the enterprise / large-city tier through a permitting beachhead, and absorbed PublicStuff (which now powers Philadelphia's 311 and, by app-store accounts, has been broken for months). Accela is high-cost and widely disliked: six-and-seven-figure deployments (Seattle spent $12.3M), steep learning curve, specialist-only configuration, long implementations, and high-profile botched launches - the Seattle permit system was called "a big disaster" in the local press. Small and mid-size jurisdictions cannot absorb that risk.
Granicus (govDelivery) - dominates outbound citizen communication (email/SMS/social mass comms, segmentation) and holds large-agency relationships. But its service-request product (govService) is far less mature than its communications products. Granicus owns proactive comms; it does not own service-request intelligence.
QScend / QAlert - a solid, call-center-oriented citizen-request manager, more usable and better-priced than Accela, now under Catalis. Dated UI, minimal modern self-service or AI.
GOgov, Comcate, Rock Solid / OneView - competent mid-market players for small-to-mid municipalities. Simple, affordable, thin on advanced workflow, analytics, and AI. Rock Solid is tied to the cost and complexity of Microsoft Dynamics.
Salesforce Public Sector Solutions - powerful and genuinely extensible with a strong security posture (Government Cloud), but it needs expensive system integrators, long timelines, and per-license cost. Overkill for a county that wants a working 311, not a platform project.
Polimorphic - the best-funded newcomer ($28M raised, $18.6M Series A). A full AI platform: conversational chat, voice AI for phone calls, an AI "answers engine," forms, CRM, workflows, permitting. It publishes hard ROI numbers (claims of 90% voicemail reduction, 75% fewer walk-ins). This is the company to beat on AI. Its relative weakness: it is strong on the AI front door (intake, deflection, answers) and thinner on the back half - closed-loop, proof-of- resolution status updates and field-crew workflow.
Citibot - AI chat and voice for resident engagement, SMS/web, 24/7, 70+ languages. Genuinely AI-forward on conversational multilingual intake. Lighter on back-office workflow and CRM depth.
The incumbents bolt a chatbot onto a forms-based CRM and call it AI. The AI-natives treat AI as the front door - conversational intake over text, voice, and phone that deflects calls and answers questions instead of just filing tickets. The AI-natives are winning new contracts because they can show ROI numbers the incumbents cannot.
Our research converged on a consistent set of failures. CivicLoop's architecture is organized to close each one.
| Failure mode | Evidence | How CivicLoop closes it |
|---|---|---|
| The "black hole" - requests vanish, no status, no proof | Chicago Inspector General found requests marked "completed" with work undone; NYC311 reviews report tickets closed within minutes, no work done; residents explicitly demand photo proof of resolution | The living record (block 3) + proactive comms engine (block 6): every status change notifies the resident; resolution requires a proof photo |
| English-only / weak language access | Multilingual jurisdictions running English-only apps; resident complaints about language barriers | 35-language voice + text intake on every surface (blocks 1 & 6) - and in PGC, this maps to county law (section 5) |
| Broken / abandoned apps | PGC311 at 1.5 stars, can't submit; PublicStuff broken for months; apps that redirect to buggy department forms | A product that actually works, demoed live, on the proven Ta-Tech engine |
| Mandatory accounts, password loops | Recurring resident complaint across systems | Report without an account - phone number or anonymous; account optional, only for tracking convenience |
| Weak analytics | Repeatedly cited for SeeClickFix and others - cities cannot see request patterns | The director dashboard (block 7): category mix, SLA performance, geographic heat-map, trend lines |
| No work-order / back-end integration | Academic case studies (NYC311, Philly311, SF311, Orange County FL) converge on this as the #1 city-side failure - the 311 CRM is not connected to departmental systems, so no real-time status | Open311 API + integration layer (block 8); the agent console is the work-order surface for departments that lack one |
| Brutal implementations / cost overruns | Accela-class failures; general government skepticism rooted in prior failed CRM projects | Built on a proven engine, deployed in a focused scope, offered as a no-cost pilot - the County risks nothing |
| Departmental turf / routing resistance | Case studies cite "liaising between the 311 center and municipal departments" as a perennial challenge | AI auto-routing with visible rationale - the routing is transparent and overridable, so departments trust it instead of fighting it |
| Surprise SLA breaches | Reactive-only systems; nobody sees the breach coming | The prediction layer (block 5): SLA-breach forecasting early enough to act |
Three capabilities exist in the market in pieces. No vendor combines all of them:
The intersection of those three, plus genuine analytics, is empty. CivicLoop takes the intersection. That is not a feature list - it is a defensible position, because the incumbents would have to rebuild their core to follow us there, and the AI-natives would have to build the back half they skipped.
Beyond the three-way intersection above, CivicLoop now carries four operational features no surveyed rival (SeeClickFix, Accela, Polimorphic, Citibot, QScend, GOgov, Comcate, Rock Solid, the rushilpatel21/CivicPulse open-source clone, or the Romulus 311 stack) ships. Each is concrete, demoable, and shipped today.
/admin, set by the county
admin: off (no automation), route (AI auto-routes to
department), full (AI auto-routes AND auto-assigns to the
lightest-loaded agent AND sends the assigned-resident
notification). Every step is logged with actor='ai' and
AUTOPILOT: prefix and is reversible. No rival lets a county
choose its own automation level on a single, governable dial.
Polimorphic delivers AI as a black box you cannot tune; the
incumbents deliver no automation at all./public, anonymous, no account,
no PII, shows last-7-day filed / resolved / open / median
resolution / SLA on-time / NPS, top categories, by-district, an
anonymized SVG scatter map, plus a "save / print PDF" button and a
CSV at /api/public/weekly.csv. SeeClickFix has a public map but
it is a marketing surface, not a governance surface; nobody else
exposes weekly SLA and NPS to the public in machine-readable form.(category, council_district),
predicted_issues holds a 7-day projection computed from a
12-week mean times recent seasonality, with a confidence score and
a director "Run forecast now" button. Polimorphic talks about
intake AI; nobody surveyed ships an outbound operational
forecast a Director can act on.These four are the differentiators we lead with in objections about "every 311 vendor says AI." We say: every vendor says it; here are four screens you can point at, today, that none of them ship.
PGC is the most populous Black-majority county in the United States: roughly 58-60% Black, about 22% Hispanic/Latino, about 25% foreign-born. The County school system already serves 20+ languages - Spanish, Amharic, Haitian Creole, Pashto, French, Yoruba, Mam, K'iche', Vietnamese, and more.
Critically, the County has a Language Access for Public Services Act (2017, amended 2020) that legally requires county agencies to provide enhanced language access.
This reframes CivicLoop's multilingual capability entirely. It is not "nice tech." The incumbent PGC311 app is English-centric and cannot even submit a request. CivicLoop's 35-language voice and text intake is the only path we are aware of by which the County's 311 function actually complies with the County's own language-access law. We pitch it as compliance, not as a feature.
These are not differentiators. They are the price of being in the room. Documents 09 and 08 specify how we meet each.
Open311 / GeoReport v2 is an open, collaborative REST API standard for civic issue reporting. It defines a common interface for viewing service definitions and submitting and tracking service requests, replacing proprietary forms and phone-only intake. It has been supported since around 2010 by San Francisco, Washington DC, Boston, and dozens of other cities, and by vendors including SeeClickFix and Connected Bits.
Committing to Open311 GeoReport v2 lets CivicLoop:
Our positioning line: open standard in, CivicLoop intelligence and closed-loop status on top. The County keeps its data and its freedom; we earn the renewal on results, not on lock-in.
Two new wedges (7 and 8) have emerged from the shipped wave, and they are the ones we close on:
off to route to full while the panel watches the
next inbound request route, assign, and notify itself in under
ten seconds. No rival lets a County dial its own automation in
public; this is the moment that ends the "is the AI real" debate.Every competitor in Section 2 sells the County a product: a reporting app, a CRM, a comms tool. CivicLoop's positioning is structurally different and harder to compete with: 311 is the front door to a County Intelligence Platform (Document 01, Section 9).
This is a moat, not a slogan:
We say it to the panel plainly: the County is not buying a 311 app. It is starting a County Intelligence Platform, and 311 is the door that is open today. That reframes the whole evaluation - from "which 311 app is best" (where incumbents have a distribution head start) to "who can give us a coordinated County" (where we are the only credible answer in the room).
Every one of the eight system building blocks in Document 01 now has a documented competitor failure it is designed to close. The build is not a list of features we think are interesting. It is a systematic shutdown of every known way a 311 system fails a county and its residents. The remaining documents - domain model, roles, architecture, screens, AI components, integrations, security - each trace back to this strategy.
Next: 03 - Domain Model - every entity, every relationship, and the lifecycle of a service request.