Tools for connectivity, security, private models, and product video
AWS appliances, self-hosted Gemma and Qwen, Laya on AWS and Azure, hosted scanning, a search API for public cloud storage, and Splainly for product video.
Network Optimization
Move traffic where it needs to go, and reach hosts that sit behind a firewall.
Multi-Port Forward Server
Listens on the ports you choose and forwards TCP or UDP to internal or external hosts. The destination port can change on the way through.
- One gateway for many listening ports
- Forward to private hosts or external addresses
- Translate the destination port as traffic is forwarded
- A direct alternative to a stack of NAT gateways
Reverse SSH Server
Persistent reverse SSH so you can reach servers and devices behind firewalls, without a public IP on each one.
- Reach private-IP hosts from one server
- Key-based authentication per client
- Clients reconnect after a drop or reboot
- Add and remove clients as the fleet changes
Cloud Security
See attacks as they happen, and scan for the issues you can fix before someone else does.
HoneyDrop
A honeypot for your VPC. Decoy services draw attention, and every interaction is logged to CloudWatch with alerts to Slack, Discord, or Teams.
- Decoys for SSH, FTP, HTTP, MySQL, Redis, RDP, Telnet, and more
- Alerts as soon as something touches a decoy
- CloudWatch logs you can wire to alarms and automation
- Deploy with CloudFormation into the subnet you choose
Panoptic Scans
Hosted vulnerability scanning for networks and web apps. Schedule OpenVAS, ZAP, Nmap, and Nuclei, then keep the reports.
- Network, port, and application scans from one place
- Schedules that match the cadence your audit expects
- Email reports, with a path into Vanta
- Plans from $25 a month
Security APIs
Look up what is already sitting in public cloud storage.
Unsecured Public Storage
Search metadata for files already public in AWS S3, Azure Blob, Google Cloud Storage, DigitalOcean Spaces, and Alibaba Object Storage.
- Filter by keyword, path, extension, size, date, provider, and bucket
- Returns filenames, sizes, last-modified times, and public URLs
- Metadata only. The API does not download file contents
- Same search on RapidAPI and Zyla Labs
AI Video
Product video from the material you already have.
Splainly
Turn a screen recording, screenshots, or a product pitch into an explainer, a sales video, or a vertical UGC short.
- A lip-synced presenter on the finished video
- A script you review before sales and UGC videos are generated
- Landscape explainers and 9:16 shorts
- New accounts start with free seconds
Private Models
Self-hosted models in your own cloud account. You pay for the instance, not for tokens.
Gemma 4 26B A4B
Gemma 4 26B, with 4 billion active parameters, on a GPU instance in your AWS account. Text and image input, with no per-token charge.
- Stays in your AWS account
- Hourly software fee, unlimited use of that instance
- Reasoning, coding, and image input
- Free trial on AWS Marketplace
Qwen 3.6 35B-A3B
Qwen 3.6 35B with 3 billion active parameters. A mixture-of-experts model for coding, reasoning, and image or video input.
- Runs as one GPU instance in your account
- 262K context, with a longer window available
- Thinking mode and a direct-response mode
- Billed by the hour, not by the token
Qwen 2.5 7B
A smaller Qwen 2.5 model for instruction following, code, and structured output, hosted on a GPU instance you control.
- 128K context
- Stronger coding and math than earlier 7B Qwen builds
- More than 29 languages
- Hourly fee, with a free trial
Qwen 3.8 27B Uncensored
A self-hosted Qwen 3.8 27B vision-language model in your AWS account. This build has refusals removed and is meant for research and controlled evaluation. Add your own safety layer before any end-user deployment.
- Runs only in your account
- 262K context, thinking mode, and tool calling
- Image understanding, including text in images
- Apache 2.0 weights, billed by the hour
Laya Decision Engine
A self-hosted decision engine on a GPU instance in your AWS account, or a GPU virtual machine in Azure. It answers typed questions and returns calibrated scores. It does not generate text.
- Choice, score, and yes-or-no questions over text, email, tickets, or JSON
- A router for English, multilingual, and typed-decisions checkpoints
- About 33 milliseconds per question on a T4
- Same engine on AWS Marketplace and Azure Marketplace
Custom Work
When the off-the-shelf product is close, but not the thing you need
Some problems still need a build of their own. We design and ship that work the same way we ship our products: small, documented, and ready for your team to run.
How a custom engagement goes:
- 1Assessment
We look at what you already run and where it is getting in the way.
- 2Solution Design
We design something that fits the constraint, with cost and operations included.
- 3Implementation
We build it in your environment and put it into service.
- 4Documentation & Training
You get the docs and a walkthrough so your team can operate it.
- 5Ongoing Support
We stay available as the setup changes.
Tell us what you are trying to run.
Send a note and we will reply with a straight answer about which product fits, or whether it needs custom work.
Get Started