Intent Detection in Email Workflows

🚨 I just tested Maia inside Make… and this might change how Ai automation/AIAgents gets built. Not templates. Not a chatbot. This felt like pair programming with Make itself. Make just released Maia (Beta) an AI builder that sits directly inside the scenario canvas and helps you build workflows through conversation. So I decided to stress test it. I gave it a real workflow: 👉 Analyze every Gmail email 👉 Classify it into Support, Sales, Spam or Finance 👉 Apply Gmail labels automatically 👉 Send Support emails to Slack 👉 Log finance emails into Airtable And within seconds Maia started building the scenario architecture on the canvas. What it generated: 📩 Gmail → Watch emails 🧠 AI → Classify email intent 🏷 Gmail → Apply label 🔀 Router → Route by category 💬 Slack → Send notifications 📊 Airtable → Log finance transactions But here’s what actually impressed me. It didn't just build modules. It explained why things were failing. For example it told me: ⚠️ If lable exists or not or it waits for you to create, confirm creation and continue building, same for slack channels, it waits for your confirmation ⚠️ Airtable record creation can't map fields without loading the table schema ⚠️ Gemini connection cannot be used directly inside the AI module or use the Make AI Provider connection instead That’s not template generation. That’s workflow reasoning. And the experience feels very different from typical automation builders. You are still seeing the full scenario: • every module • every route • every connection • every decision Nothing becomes a black box. From what I tested today, Maia already seems capable of building a big portion of basic to intermediate automations automatically. Think: ⚡ Email routing ⚡ AI classification workflows ⚡ Slack alert systems ⚡ Airtable logging pipelines ⚡ CRM triage flows Which previously required someone who knew Make well. But the bigger shift here is something else. We’re moving from: Automation → AI Automation → AI Systems And Make is quietly becoming an AI orchestration layer for business workflows. Because now you have: 🧠 AI agents 🔌 APIs 🧰 tools 🧩 routers, If/else - Merge 🤖 AI decision making All inside the same canvas. If you understand systems thinking and process design, tools like Maia will make building workflows 10x faster. Still testing the limits of this. But first impression? This is one of the most interesting things Make has shipped recently. Curious if others here have tried Maia yet. What workflows did you test? #AI #Automation #Make #AIagents #iPaaS #BuildInPublic

Recently, a client reached out to us expressing frustration with the RAG (Retrieval-Augmented Generation) application they had implemented for customer support emails by a different AI agency. Despite high hopes of increased efficiency, they were facing some significant problems: The RAG model frequently provided wrong answers by pulling information from the wrong types of emails. For example, it would respond to a refund request email with details about changing an order - simply because those emails contained some similar wording. Instead of properly classifying the emails by type and intent, it seemed to just perform a broad embedding search across all emails. This created a confusing mess where customers were receiving completely irrelevant and nonsensical responses to their inquiries. Rather than streamlining operations, the RAG implementation was actually making customer service much worse and more time-consuming for agents. The client's team had tried tuning the model parameters and changing the training data, but couldn't get the RAG application to accurately distinguish between different contexts and email types. They asked us to take a look and help get their system operating reliably. After analyzing their setup, we identified a few key issues that were derailing the RAG performance: Lack of dedicated email type classification The RAG model needed an initial step to explicitly classify the email into categories like refund, order change, technical support, etc. This intent signal could then better focus the retrieval and generation steps. Noisy, inconsistent training data The client's original training set contained a mix of incomplete email threads, mislabeled samples, and inconsistent formats. This made it very difficult for the model to learn canonical patterns. Retrieval without context filtering The retrieval stage wasn't incorporating any context about the classified email type to filter and rank relevant information sources. It simply did a broad embedding search. To address these problems, we took the following steps with the client: Implemented a new hierarchical classification model to categorize emails before passing them to the RAG pipeline Cleaned and expanded the training data based on properly labeled, coherent email conversations Added filtered retrieval based on the email type classification signal Performed further finetuning rounds with the augmented training set After deploying this updated system, we saw an immediate improvement in the RAG application's response quality and relevance. Customers finally started getting on-point information addressing their specific requests and issues. The client's support team also reported a significant boost in productivity. With accurate, contextual draft responses provided by the RAG model, they could better focus on personalizing and clarifying the text - not starting responses completely from scratch.

How AI SDRs actually work? AI SDRs do more than just send emails—they: - identify leads - analyze intent - personalize outreach - qualify prospects - book meetings all at scale. Here’s how they work And the tools that power them: 1️⃣ Lead Identification & Data Enrichment: AI finds and enriches leads using: RB2B → Identifies anonymous website visitors. Breakcold → Checks CRM to avoid duplicate outreach. Trigify.io & Leadspicker → Tracks job changes, funding rounds, and hiring trends. AI scans LinkedIn, job boards, and databases to find buyers Adds context like company news, and ensures outreach is fresh. 2️⃣ Understanding the Prompt & Generating Outreach: AI SDRs don’t just send generic messages. They follow structured prompts that define: - Goal (book a call, follow up, gather info). - Lead details (company, role, activity, pain points). - Tone & personalization (casual, direct, professional). AI also retrieves past conversations to maintain context. 3️⃣ Prioritization & Intent Detection: Not every lead is worth chasing! AI qualifies and ranks prospects based on engagement. How It works: - AI analyzes email opens, LinkedIn engagement, and CRM signals to score intent. - High-intent leads move to a high-touch sequence with immediate follow-ups. - Unresponsive leads are dropped or nurtured passively to avoid wasting time. 4️⃣ Handling Conversations & Lead Qualification: AI SDRs respond over email, chat, or voice and qualify leads using: Humanlinker & Amplemarket → Video and voice note-based LinkedIn outreach. BANT framework: to assess Budget, Authority, Need, and Timeline. If a lead fits, AI books a meeting or routes it to a human SDR. 5️⃣ Automated Follow-Ups & Smart Nurturing: Follow-ups are adjusted based on engagement: HeyReach → LinkedIn DMs. Drippi.ai → Twitter outreach. Cold DM → Instagram messaging. Smartlead / Instantly.ai → Cold email AI changes messaging angle and outreach channel (email → LinkedIn → Twitter) to improve response rates. 6️⃣ CRM Integration & Learning: Persana AI / Airscale / Clay → Optimize targeting and refine outreach. Tracks what’s working and adjusts messaging automatically. 7️⃣ Multichannel Outreach: If email fails, AI shifts to another channel where the lead is active. They find, engage, and qualify leads automatically, Allowing human reps to focus on closing deals. If you're not using AI for outbound, you're already behind. #ai #aisales

We’ve had over 80,000+ folks check out Architect. The single most requested integration? Gmail. Not “write emails for me.” People want to automate decision-making inside their inbox. Here are the top 10 Gmail use cases we’re seeing: 1) Thread Summaries: Condense 15+ reply chains into decisions, open questions, and next steps. 2) Deal Signal Detection: Flag urgency, pricing intent, churn risk, or buying signals automatically. 3) Context-Aware Drafts: Generate replies that pull CRM history, past conversations, and relevant documents. 4) Workflow Triggers: “Contract signed” → notify finance. “Book demo” → create meeting + update CRM. 5) Auto CRM Updates: Log activity, update deals, append notes, no manual admin. 6) Smart Follow-Ups: Detect no-response windows and adjust tone or escalate automatically. 7) Priority Classification: Categorize emails (Revenue, Hiring, Legal, Support) and highlight high-impact ones. 8) Executive Briefings: Daily/weekly summaries of critical threads and revenue-impact conversations. 9)Attachment Processing: Extract contracts, parse invoices, identify missing clauses from PDFs. 10) Multi-Agent Inbox: Sales agent. Support agent. Finance agent. Each reading the inbox with domain context. The pattern across the users is clear: People don’t want AI to “help write emails.” They want AI to remove friction, extract signal, and trigger action. Your inbox is an intelligence surface. Most teams just haven’t automated it yet. Build your Gmail automation today in < 5 Minutes. #architect

Every missed intent signal costs you ~$2K in lost pipeline per lead. Here’s how we fixed that. Most teams think their outbound problem is “low conversions.” It’s not. It’s missed intent. Because while reps chase cold leads, real buyers are already showing signals and nobody’s watching. At DevCommX, we built a full Intent Signal Engine that connects data across 3 layers: 🔹 1️⃣ Direct Insight Triggers We track 9+ live buyer signals - job openings, funding rounds, tech changes, and firmographics - using tools like Clay, #Apollo, BuiltWith, and Similarweb. ▫️ Result: Our reps now engage 11–13 days before competitors do. 🔹 2️⃣ Partner Data Signals We plug into review sites, ABM platforms, and affinity data. So when your buyer compares vendors, you already know. ▫️ Result: +36% faster deal acceleration. 🔹 3️⃣ External Data Signals From product usage to website visitors and LinkedIn engagements, every digital footprint gets scored for intent and timing. ▫️ Result: +44% more qualified pipeline in 6 weeks. We’re not guessing anymore. We’re reading intent in real time - and that changed everything. 🔸 Pro tip: Don’t add more leads. Add more context. That’s how pipeline compounds. ▫️ If you want the exact intent stack and signal mapping we use, Comment “INTENT” below - and I’ll send over our Buyer Intent Workflow Map (the same one in this post). #RevOps #IntentData #OutboundAutomation #SalesIntelligence