What we're building, and what we won't.
Most companies hide their roadmap behind a sales call. We'd rather show you exactly where Klaros is headed so you can decide if it's the right fit, today or in three months.
Last updated: 13 August 2026 · See what shipped → · Request a feature →
Already shipped
Live in productionThese are in your hands today, not "coming soon," not behind a flag.
Now: actively building
In developmentCode is being written. These will ship within weeks, not months.
CTWA rollup analytics
Aggregate Click-to-WhatsApp ad performance into Overview and Costs dashboards: cost-per-conversation from ad spend through to reply, broken down by ad campaign.
Relationship-stage progression
Contact rows in the inbox gain a relationship-stage chip: from first-touch through active conversation to retained. The stage derives from consent state, reply cadence, and conversation history. Filters and lenses let you focus on contacts at each stage.
Visual flow builder
Drag-and-drop conversation flow editor: if/then branching, wait nodes, API calls, human handoff. The BSP flagship feature, though our developer buyer often prefers code.
Next: planned, not started
PlannedDesigned and scoped. Will begin after the "Now" items ship. Roughly 2–4 months out.
Threads that need a person, surfaced before they rot
The groundwork shipped in August: Klaros already detects a conversation where the customer spoke last and nobody replied, works out how long is too long from that business's own median reply time, and alerts the operator. On its first run it found a real enquiry that had been waiting 62 hours. What's next is turning that same detection into a short, ranked list of relationships that need attention today, each one carrying the reason it is on the list.
Conversation memory that survives the context window
Today the assistant reads the most recent stretch of a thread. That is a window, and a window is a truncation: on a long relationship it cannot know how the relationship began. The plan is to summarise a conversation when it ends rather than discard it, so a five-year customer costs the same to answer as a stranger and is answered better.
Four more vertical packs
Legal Firm, Fleet & Logistics, Facility Management, and Healthcare Clinic join the six packs already live in the Extensions catalog. Same entity engine, same watchdogs and digests, a new domain configuration each.
Portfolio dashboard
A single screen that shows every active entity (project, matter, site): status, last update, team responsiveness, flagged interventions. Built for the person who manages 10 projects and currently scrolls 10 WhatsApp groups every morning.
AI extraction: photos, voice, documents
When someone sends a site photo, a voice note, or a PDF report, the system reads it. A photo gets classified ("rebar installation, level 3, appears complete"). A voice note gets transcribed and tagged. A progress report gets its key numbers extracted. All filed under the right entity, automatically.
AI next-best-action
Graph-driven suggestions: "Priya hasn't replied in 7 days: follow up?", "Rahul overlaps with 3 other contacts: merge?", "This segment's reply rate dropped 40%: pause campaign?" Powered by Workers AI on the consented relationship graph.
Cross-entity performance scorecards
When the same contractor, vendor, or team member works across multiple projects, their performance pattern becomes visible: responsiveness, issue frequency, delay correlation. The data already flows through WhatsApp. We just connect the dots.
Agent performance + SLA
Response time tracking, resolution metrics, SLA timers, working hours configuration. Agent leaderboard and bottleneck detection. Depends on the multi-user model shipping first.
Published Zapier / Make app
A listed app in the Zapier and Make directories, so connecting Klaros is a search-and-click instead of pasting in a webhook URL. The integration path itself already works today via each platform's generic HTTP/webhook modules and our OpenAPI spec (see "n8n / Zapier / Make integration path" above). This is the packaging layer on top of that, not new plumbing.
Later: exploring
3–6 monthsOn the radar, not committed. We'll build these when a paying customer pulls, not when a roadmap says we should.
Mobile PWA inbox
A progressive web app that works on your phone: reply to conversations, see notifications, check costs. Not a native app (yet), but close enough for most teams.
Commerce: recurring orders + native checkout
Recurring orders for repeat-purchase verticals (jewellery, FMCG), plus a native in-chat checkout (WhatsApp Pay / UPI) that doesn't need a payment link at all. Table stakes for commerce-first use cases.
FAQ AI agent
An AI agent grounded in your knowledge base that handles tier-1 questions automatically; escalates to a human when it's unsure. Powered by Workers AI.
Shopify / WooCommerce app
Order event notifications (confirmed, shipped, delivered), abandoned cart recovery, product feed sync. Built on top of catalog + webhook infrastructure.
Not planned: by design
Won't buildThese are deliberate choices, not oversights. Knowing what we won't build is as important as knowing what we will.
Instagram / Facebook / SMS / multichannel inbox
Klaros is WhatsApp-only. We don't believe in "unified inboxes" that do everything poorly. WhatsApp is the highest-ROI messaging channel in our markets. We go deep on one channel instead of shallow across five.
Per-message markup or hidden fees
We will never add a per-message markup. Our code literally has markup: 0. Revenue comes from the annual software license. Your messaging cost is Meta's rate, period. This is load-bearing architecture, not a marketing promise.
Hosted / multi-tenant SaaS
We won't host your data. Klaros runs on your Cloudflare account because data sovereignty is the point. If you want a vendor holding your customer conversations, there are plenty of BSPs. That's not what we are.
Enterprise RBAC / SSO / department routing
We ship auto-assignment (round-robin, least-busy, sticky, skill-based) and per-agent capacity caps. What we do not build: Salesforce-grade role hierarchies, department-level queue routing, or SSO federation for 50+ agent call-center teams. The line is practical team routing versus enterprise ticket infrastructure.
Becoming a payment processor
We ship the order tracking and the in-chat payment request (India, WhatsApp's order_details message), but the payment link itself comes from your own provider (Razorpay, PayU, whoever you already use), and the money never touches us. We won't become a PSP or hold funds. That's market-specific, tightly regulated, and orthogonal to our value prop.
How we decide what to build
Pull, not push
We don't build features because a competitor has them. We build when a paying customer needs something and can't work around it. Competitor parity is a losing race.
Wedge over parity
Features that deepen our moat (cost transparency, consent, relationship context, self-deployed sovereignty) get priority over features that just match what BSPs already have.
Depth over breadth
We'd rather be the best WhatsApp platform than a mediocre omnichannel tool. One channel, done right, beats five channels done half-way.
Ship, then polish
Every feature on this page ships as the minimum useful version first. We iterate based on real usage, not guesses about what "complete" looks like.
Get notified when features ship
Drop your email. We'll ping you when the features you care about go live. No spam, no drip sequences (ironic, we know).
