Our product · In development
Upskill News: keep every client moving between check-ins.
A personalized newsletter and podcast platform for independent educational consultants. It's built to do the research legwork — verified, sourced and labeled — so the consultant adds what software never can: judgment, context and a personal note.
Upskill News is a working prototype in active development. We're opening conversations with IECs who want to shape it — no pricing, no launch date promises, just an honest look at what we're building.
The idea
Aiming to handle ~90% of the research, so you add the judgment.
Between meetings, deadlines shift, programs update, and visa rules change. Upskill News is designed so each client gets a personalized briefing on their schools and programs — every factual claim checked and cited — while the consultant stays firmly in charge of what actually goes out.
How it works
Four steps, with a human in the loop by design.
Configure each client
Schools of interest, program tags, delivery cadence (biweekly or monthly), newsletter, podcast or both, tone preference — set once, refine anytime.
Verification-first research
Every factual claim is labeled Verified, Stale or Unverified, with a source link and retrieval date. Sources that disagree are flagged, never guessed. Freshness rules keep time-sensitive facts like deadlines on a short leash.
You review and approve
Approval is blocked until every flagged claim is individually confirmed by the consultant. Add your own "tidbits" — personal notes that render in a distinct from-your-consultant section. Nothing ships without your sign-off.
Deliver where clients live
Designed to deliver a scannable newsletter and an in-app podcast player, wrapped in a mobile-first installable web app — no app store required.
Built for trust
Skeptical software, by design.
The whole product is organized around one belief: AI-assisted research is only useful to a consultant if it shows its work and keeps a human in charge.
Confidence tiers on every fact
Verified, Stale or Unverified — each claim carries its source and retrieval date, or an explicit "generated from model knowledge" label. Stale or disputed facts are kept out of the spoken podcast entirely.
Consultant approval gate
Clean editions approve instantly; flagged claims each require an explicit "I've verified this myself." Friction is proportional to risk, and editorial responsibility stays with you.
Platform quality gate
Editions with risk signals — source conflicts, stale deadlines, low verified share — are automatically held for human review before they ever reach the consultant. A held edition physically cannot be sent.
Newsroom-style corrections
If something delivered turns out wrong, the platform issues a labeled correction notice and banners the edition. Originals are never silently rewritten, and a platform-wide corrections log keeps the record.
Consultant tidbits, private by design
Your personal notes render in their own section of the newsletter — and are deliberately never sent to the voice-generation service, so they stay out of the podcast audio.
Consent-first delivery
Designed around a double opt-in email consent model with an auditable trail; unconsented sends are hard-blocked inside the delivery logic itself.
A client experience worth opening
Podcast player with skips, speed and position memory; timestamped show notes with source links; a past-editions archive; PDF export. Installable as an app, with offline reading planned into the architecture.
Accessible and tested
Automated WCAG 2.2 AA audits across the app report zero violations, and 640+ automated tests plus append-only audit trails back the engineering. Real assistive-technology user testing is still ahead of us.
A look inside
The product today — real screens, fictional data.
These are actual screenshots of the working prototype. Every name, school, deadline, statistic and activity metric shown is fictional demo data — nothing here is real client or usage information. The app watermarks demo content so it can never be mistaken for real guidance.
The consultant dashboard: upcoming deliveries, tidbits due, and client engagement at a glance. All client names and activity shown are fictional personas. 
Edition review: every claim carries a confidence tier, source and retrieval date before you approve. 
The platform quality gate: risky editions are held for human review before reaching the consultant. The held-rate and verified-share figures are simulated demo telemetry, not real operations data. 
The client view: podcast player, show notes — and honest, labeled corrections when something changes.
Two minutes in the consultant workspace
A short walkthrough of the working prototype — configuring a client, reviewing a fact-checked edition and approving it for delivery. Everything on screen is fictional demo data.
Where it stands
What's real today — and what's next.
We'd rather under-promise. Here is the honest state of the product as of mid-2026.
Working today
- A functional prototype: client configuration, edition review with confidence tiers, the consultant approval gate, the platform release gate, corrections workflow, and the client reading experience — including the in-app podcast player, though narrated audio itself is still ahead.
- A verification-first content pipeline with per-fact sourcing, freshness rules and QA checks, backed by 640+ automated tests.
- Mobile-first installable web app; automated WCAG 2.2 AA accessibility audits report zero violations.
Still ahead
- Production infrastructure: live database, real accounts and scheduled hands-off delivery are planned next.
- Narrated podcast audio, live email delivery at scale, and the full client sign-up flow.
- Pricing and general availability — neither exists yet, and we won't announce dates we can't keep.
Are you an IEC? Help us build this right.
We're looking for independent educational consultants to pressure-test Upskill News as it develops. Tell us how you keep clients informed today — we'll show you what we're building and where it's headed.