The latest articles on Forem by Aikit Pros (@aikitpros). https://forem.com/aikitpros https://media2.dev.to/dynamic/image/width=90,height=90,fit=cover,gravity=auto,format=auto/https:%2F%2Fdev-to-uploads.s3.amazonaws.com%2Fuploads%2Fuser%2Fprofile_image%2F3891008%2F00c7c6eb-42dd-4861-af3c-6517f75c0d6c.jpg https://forem.com/aikitpros en Aikit Pros Sat, 25 Apr 2026 16:33:30 +0000 https://forem.com/aikitpros/building-aikitpros-orchestrating-7-gen-ai-apis-behind-one-brief-55aj https://forem.com/aikitpros/building-aikitpros-orchestrating-7-gen-ai-apis-behind-one-brief-55aj <h2> TL;DR </h2> <p>I built <a href="https://aikitpros.com" rel="noopener noreferrer">AikitPros</a> — a creative hub that takes a single brief and orchestrates Midjourney, Flux, Suno, Sora, Luma, GPT and Claude into one campaign output (script + images + music + video). Total cost in production: $0.35 per campaign. Live demo: <a href="https://x.com/aikitpros/status/2046596943023890780" rel="noopener noreferrer">https://x.com/aikitpros/status/2046596943023890780</a></p> <p>This post is the architectural walk-through.</p> <h2> The problem </h2> <p>Most "all-in-one AI" tools are thin wrappers that just expose 7 buttons. The real value is coordination: the music tempo has to match the video cut, the on-screen copy has to fit the image composition, the voice-over script has to land in 8 seconds. Calling 7 APIs in parallel and concatenating the outputs gets you ~40% usable rate. Not good enough.</p> <h2> Architecture </h2> <p>Brief -&gt; Planner (Claude) produces script, shot list, audio direction, brand constraints. Then a fan-out routes to ImageRouter (Midjourney V8 / Flux / GPT Image 2), MusicRouter (Suno), VideoRouter (Sora / Luma / Hailuo), and CopyRouter (GPT + Claude). A judge model (GPT-4o-mini) reviews the assembled output and either delivers it or re-runs only the failing modality.</p> <p>The judge step is the one decision that took 40% usable to 90%+ usable.</p> <h2> Three things I would do differently next time </h2> <ol> <li>Cache the planner output. Same brief = same plan. I burned a lot of Claude tokens regenerating identical decompositions.</li> <li>Stream partial results. Users tolerate 60s of waiting if they see the script appear at 5s, the image at 20s, the music at 35s.</li> <li>Pre-warm the video model. Cold starts on Sora/Luma can add 30s. A keep-alive ping during the image step parallelizes that latency away.</li> </ol> <h2> Try it </h2> <p>Hub is live at <a href="https://aikitpros.com" rel="noopener noreferrer">https://aikitpros.com</a>. First credits are on me — if you build something with it, I would love to see it.</p> <p>If you are building something similar and want to compare notes on judge models or routing heuristics, drop a comment.</p> ai webdev Aikit Pros Fri, 24 Apr 2026 02:39:39 +0000 https://forem.com/aikitpros/one-subscription-20-ais-how-i-built-aikitpros-to-replace-my-55mo-ai-stack-35nl https://forem.com/aikitpros/one-subscription-20-ais-how-i-built-aikitpros-to-replace-my-55mo-ai-stack-35nl <h2> What if your entire 30-second ad campaign cost $0.41? </h2> <p>Not the fee for the tools. The TOTAL cost — script, hero image, custom soundtrack, rendered video, and platform copy.</p> <p>I just spent 14 days replacing my $55/mo AI subscription stack with a single workflow built on top of <a href="https://aikitpros.com" rel="noopener noreferrer">AIKitPros</a>. Here's the stack, the numbers, and the retry pattern that took my failure rate from 20% to under 2%.</p> <h2> The problem </h2> <p>I was paying for ChatGPT Plus, Midjourney, ElevenLabs, Runway, and more — about <strong>$55/month</strong> — just to produce a single short-form campaign. Each tool sat in its own silo. Switching context cost me hours.</p> <h2> The stack </h2> <p>One unified API hub exposing 20+ models behind a single token:</p> <ul> <li> <strong>Script &amp; copy</strong>: GPT-class text models</li> <li> <strong>Hero image</strong>: Flux / SDXL</li> <li> <strong>Soundtrack</strong>: Suno-class audio</li> <li> <strong>Video</strong>: Veo / Kling</li> <li> <strong>Orchestration</strong>: Dify workflow → AIKitPros API</li> </ul> <p>Total per 30-second campaign: <strong>$0.41</strong> in raw API costs.</p> <h2> The retry pattern </h2> <p>The single biggest unlock was a deterministic retry wrapper around each generation step with:</p> <ol> <li>Exponential backoff on 429 / 5xx</li> <li>Prompt mutation on content-policy rejections</li> <li>Model fallback (primary → secondary) on timeout</li> </ol> <p>That took failure rate from ~20% to under 2%.</p> <h2> Links </h2> <ul> <li>Full write-up: <a href="https://medium.com/@yujunjie1983/one-subscription-20-ais-how-i-built-aikitpros-41c6ac624fad" rel="noopener noreferrer">https://medium.com/@yujunjie1983/one-subscription-20-ais-how-i-built-aikitpros-41c6ac624fad</a> </li> <li>Live demo: <a href="https://x.com/aikitpros/status/2046596943023890780" rel="noopener noreferrer">https://x.com/aikitpros/status/2046596943023890780</a> </li> <li>Try it: <a href="https://aikitpros.com" rel="noopener noreferrer">https://aikitpros.com</a> </li> </ul> <p>Happy to answer questions about the Dify workflow, the retry pattern, or the cost breakdown.</p> ai webdev opensource tutorial Aikit Pros Tue, 21 Apr 2026 16:25:47 +0000 https://forem.com/aikitpros/how-i-chained-midjourney-suno-v4-and-veo-31-in-one-dify-workflow-to-ship-8s-video-ads-for-35-cents-5c1k https://forem.com/aikitpros/how-i-chained-midjourney-suno-v4-and-veo-31-in-one-dify-workflow-to-ship-8s-video-ads-for-35-cents-5c1k <blockquote> <p>TL;DR: one Dify workflow, four models, single API gateway. A product brief goes in, a full 8-second video ad (script + hero image + 29s soundtrack + 8s video) comes out in under 60 seconds. Real cost per run: about 35 cents. Live 8s demo + credit math on X: <a href="https://x.com/aikitpros/status/2046596943023890780" rel="noopener noreferrer">https://x.com/aikitpros/status/2046596943023890780</a></p> </blockquote> <h2> The problem </h2> <p>I run marketing for a B2B night-vision optics brand (Middle East distributors). Every regional distributor pitch clip used to mean videographer + studio + licensed music, ~$1,200 per clip, 5 days of turnaround.</p> <p>Last week I shipped 18 regional variants for under $7 total. Same pipeline, swap the brief.</p> <h2> The stack </h2> <div class="table-wrapper-paragraph"><table> <thead> <tr> <th>Node</th> <th>Model</th> <th>Credits</th> <th>Slowest-path time</th> </tr> </thead> <tbody> <tr> <td>Script + copy</td> <td>Claude 3.7 Sonnet</td> <td>~0.00</td> <td>3s</td> </tr> <tr> <td>Hero image</td> <td>Midjourney</td> <td>0.39</td> <td>12s</td> </tr> <tr> <td>29s soundtrack</td> <td>Suno v4</td> <td>0.55</td> <td>40s</td> </tr> <tr> <td>8s video</td> <td>Veo 3.1-fast</td> <td>1.53</td> <td>55s</td> </tr> <tr> <td>Total</td> <td></td> <td>2.47</td> <td>&lt;60s wall-clock</td> </tr> </tbody> </table></div> <p>2.47 credits x $0.14 = ~$0.35 per full run.</p> <h2> Why one workflow changes the math </h2> <p>Before: Midjourney Pro $10 + Suno Pro $10 + Luma $15 + ChatGPT Plus $20 = $55/mo idle for a brand that ships 4-8 clips a month.</p> <p>After: pay-per-run, one key, one bill. 18 clips = $6.30. 99% reduction vs the old freelancer spend, ~90% vs the four-subs stack.</p> <h2> Architecture </h2> <p>Dify handles orchestration. The script node finishes first, then image / music / video fan out in parallel. Wall-clock stays under 60s because the three slow nodes overlap instead of running serially.</p> <h2> The unlock: one API gateway </h2> <p>The real breakthrough wasn't picking better models. It was putting all four behind one key, one bill via Ace Data Cloud:</p> <ul> <li>No per-provider OAuth dance</li> <li>No four separate billing portals</li> <li>Credits roll across script / image / music / video</li> <li>Agentic top-up via x402 for batch runs overnight</li> </ul> <p>That's what takes the pipeline from a fun weekend hack to production infrastructure my day-job brand actually pays for.</p> <h2> Prompt patterns that actually worked </h2> <p>Script node (Claude) - force a structured shot-list JSON output so the downstream image / video nodes can key off indexed shots.</p> <p>Image node (Midjourney) - always derive the prompt from the script JSON, never from the raw brief. Lock aspect ratio and forbid on-image text.</p> <p>Music node (Suno v4) - map visual_style to BPM + mood tags explicitly. Skip lyrics; just background score.</p> <p>Video node (Veo 3.1-fast) - anchor the shot description to the hero image and forbid text overlays.</p> <h2> Five things I would love feedback on </h2> <ol> <li>Is sub-60s latency a real moat, or just nice-to-have?</li> <li>Credits-per-run vs monthly subs - does this actually change creator behaviour?</li> <li>Would you trust an agent to auto-top-up credits (x402) for batch runs?</li> <li>Weakest link in the pipeline for you - script, image, music, or video?</li> <li>Which vertical would you throw this at first?</li> </ol> <h2> Try it </h2> <ul> <li>Live product: <a href="https://aikitpros.com" rel="noopener noreferrer">https://aikitpros.com</a> </li> <li>8s demo + full credit breakdown on X: <a href="https://x.com/aikitpros/status/2046596943023890780" rel="noopener noreferrer">https://x.com/aikitpros/status/2046596943023890780</a> </li> <li>Built on Ace Data Cloud + Dify + Nexior.</li> </ul> <p>If you want the raw Dify DSL to fork, drop a comment and I'll DM it over.</p> ai tutorial showdev webdev